1
00:00:00,000 --> 00:00:02,619
Kent Beck actually said after he
saw like Gen AI for the first 

2
00:00:02,619 --> 00:00:05,770
time, he realized that 99% of 
what he knew became irrelevant, 

3
00:00:05,770 --> 00:00:09,388
but that 10% that he had learned
became a thousand times more 

4
00:00:09,388 --> 00:00:10,935
important. 
There's a danger sometimes, even

5
00:00:10,935 --> 00:00:13,688
if you are pairing two persons, 
you would still outsource your 

6
00:00:13,688 --> 00:00:16,165
thinking a lot to the AI. 
I definitely think there's a 

7
00:00:16,165 --> 00:00:18,638
danger in doing this. 
AI, at least the way they're 

8
00:00:18,638 --> 00:00:20,430
currently trained today, they 
love producing code. 

9
00:00:20,490 --> 00:00:23,282
Most of that code sometimes 
without the right guardrails, is

10
00:00:23,282 --> 00:00:25,296
gonna be gibberish. 
The next programmer who comes 

11
00:00:25,296 --> 00:00:27,720
and looks at it, they're gonna 
go, why are we doing this? 

12
00:00:27,720 --> 00:00:30,954
Code is read more by humans. 
It's meant to be read by humans 

13
00:00:30,954 --> 00:00:33,060
more than it is meant to be run 
by the machine. 

14
00:00:33,150 --> 00:00:34,450
Really what you're trying to 
design is for human. 

15
00:00:34,530 --> 00:00:37,548
And that aspect still applies. 
This week on the Tech Lead 

16
00:00:37,548 --> 00:00:40,404
Journal, meet Mohan Krishnan, a 
transformational leader who has 

17
00:00:40,404 --> 00:00:43,619
transformed engineering teams at
Grab, Bukalapak, and BBM Emtek. 

18
00:00:43,769 --> 00:00:46,680
How do you actually think about 
making a successful 

19
00:00:46,680 --> 00:00:47,769
transformational engineering 
teams? 

20
00:00:48,030 --> 00:00:51,642
My philosophy has always been 
that you wouldn't get trained by

21
00:00:51,642 --> 00:00:54,832
a tennis coach that doesn't 
himself play or herself play 

22
00:00:54,832 --> 00:00:56,730
tennis. 
I think that applies to 

23
00:00:56,730 --> 00:00:59,059
technical management as well. 
High performing engineering 

24
00:00:59,059 --> 00:01:02,140
teams are very similar to high 
performing sports teams. 

25
00:01:02,199 --> 00:01:04,818
All of this needs to be 
supported by a foundation of 

26
00:01:04,818 --> 00:01:07,406
like some level of discipline, 
which is not something we 

27
00:01:07,406 --> 00:01:09,310
normally associate with our line
of work. 

28
00:01:09,460 --> 00:01:12,320
When I see tech forums within 
this region, people are now 

29
00:01:12,320 --> 00:01:13,770
questioning should I still be in
tech? 

30
00:01:13,770 --> 00:01:15,924
I know there's a lot more focus,
are we getting rid of juniors 

31
00:01:15,924 --> 00:01:18,119
and all that? 
I don't fully buy into that. 

32
00:01:18,119 --> 00:01:20,940
I think what we mean we are 
raising the bar of what a junior

33
00:01:20,940 --> 00:01:24,120
engineer should be capable of. 
Do you think the Southeast Asian

34
00:01:24,120 --> 00:01:25,800
engineering talents are ready 
for this? 

35
00:01:41,946 --> 00:01:44,651
Hey, Mohan. 
Welcome to my next in-person 

36
00:01:44,651 --> 00:01:47,029
podcast recording. 
So you know, I've been starting 

37
00:01:47,029 --> 00:01:49,970
to doing this for tech leaders 
in at least this region, 

38
00:01:49,970 --> 00:01:53,424
Southeast Asia, first. 
So I'm very happy to have you. 

39
00:01:53,633 --> 00:01:56,177
It's been a long time coming, I 
think I've always been wanting 

40
00:01:56,177 --> 00:01:58,653
to invite you. 
And so yeah, here we are. 

41
00:01:58,773 --> 00:02:01,045
So thanks for doing this. 
No, thank you Henry. 

42
00:02:01,045 --> 00:02:04,741
I, I'm really appreciative of 
you, uh, choosing me to be on 

43
00:02:04,741 --> 00:02:06,865
your podcast. 
I hope I'll be able to share 

44
00:02:06,865 --> 00:02:07,900
something useful to your 
listeners. 

45
00:02:08,068 --> 00:02:10,001
Right. 
So Mohan, maybe in the 

46
00:02:10,001 --> 00:02:12,544
beginning, I always love to ask 
my guests first to share a 

47
00:02:12,544 --> 00:02:14,414
little bit more about yourself, 
right? 

48
00:02:14,414 --> 00:02:18,482
What type of leader you are. 
And maybe looking back from your

49
00:02:18,482 --> 00:02:21,610
career, any highlights or, I 
don't know, turning points that 

50
00:02:21,610 --> 00:02:26,559
shape you as a leader today? 
Yeah, I mean I think there are 

51
00:02:26,559 --> 00:02:30,404
definitely a couple of career 
pivotal points, that kinda 

52
00:02:30,404 --> 00:02:35,418
shaped me into both, like maybe 
my appreciation for engineering 

53
00:02:35,418 --> 00:02:39,433
and yeah, I guess you could say 
my leadership approach as well. 

54
00:02:39,992 --> 00:02:43,598
The first was definitely, you 
know, taking the plunge to join 

55
00:02:43,598 --> 00:02:47,257
Pivotal Labs in Singapore in 
2011-2012. 

56
00:02:47,617 --> 00:02:50,668
I had been working for a 
Malaysian startup up to then, 

57
00:02:50,668 --> 00:02:52,752
that was focused more on the 
logistics space. 

58
00:02:52,776 --> 00:02:55,281
And it was good and I was 
running a small team and all 

59
00:02:55,281 --> 00:02:58,041
that. 
But a lot of the things that I 

60
00:02:58,041 --> 00:02:59,603
had learned were very much 
southland. 

61
00:03:00,308 --> 00:03:02,888
Like, so we kind of read stuff, 
we try to practice it. 

62
00:03:03,468 --> 00:03:06,774
Pivotal Labs was the first 
organization that I joined where

63
00:03:06,774 --> 00:03:10,288
I learned a lot from. 
I'm not sure how much people 

64
00:03:10,288 --> 00:03:12,464
know about it, but like, so 
Pivotal Labs are in San 

65
00:03:12,464 --> 00:03:15,169
Francisco and then they, at some
point they had set up a office 

66
00:03:15,169 --> 00:03:18,725
in Singapore that was working 
very much in the same manner, 

67
00:03:18,725 --> 00:03:22,271
working with startups. 
And they are an XP shop, extreme

68
00:03:22,271 --> 00:03:26,392
programming shop, true and true.
And experiencing, I had read 

69
00:03:26,392 --> 00:03:29,962
Kent Beck's books on extreme 
programming before that, right? 

70
00:03:29,962 --> 00:03:31,942
It came out around there so late
nineties. 

71
00:03:32,392 --> 00:03:33,952
So, you know, I picked it up 
from the shelf. 

72
00:03:33,952 --> 00:03:36,877
A lot of it looked like, wow, 
can you actually do this, pair 

73
00:03:36,877 --> 00:03:38,902
programming all the time, test 
driven development. 

74
00:03:39,532 --> 00:03:41,718
And then I joined this 
organization, which pretty much 

75
00:03:41,718 --> 00:03:44,291
lived and breath it on a 
day-to-day basis. 

76
00:03:44,291 --> 00:03:46,631
And it was amazing. 
Uh, I learned a lot there. 

77
00:03:47,171 --> 00:03:49,181
So that was definitely a pivotal
moment. 

78
00:03:49,181 --> 00:03:51,633
Happy to talk through some of 
the stuff that I learned at that

79
00:03:51,633 --> 00:03:54,791
point. 
But very quickly, then I did 

80
00:03:54,791 --> 00:03:58,412
another pivotal inflection point
in my life was taking the plunge

81
00:03:58,412 --> 00:04:00,899
to move to Indonesia. 
I worked for a startup. 

82
00:04:00,899 --> 00:04:03,392
I did that with a close buddy of
mine from Pivotal Labs as well, 

83
00:04:03,392 --> 00:04:05,432
Tommy Sullivan, who's still 
there. 

84
00:04:05,642 --> 00:04:07,551
Maybe one day you'll have a chat
with him. 

85
00:04:07,769 --> 00:04:10,574
And that was really an 
experience as well where we had 

86
00:04:10,574 --> 00:04:13,667
the opportunity of taking a lot 
of stuff we learned in Pivotal 

87
00:04:13,667 --> 00:04:16,067
Labs and trying to practice in a
completely different culture. 

88
00:04:16,096 --> 00:04:19,307
I mean, completely different is 
probably overstating it, but uh,

89
00:04:19,337 --> 00:04:21,647
Indonesian culture is different,
right? 

90
00:04:21,928 --> 00:04:24,449
The Southeast Asian countries 
have their own distinct flavor, 

91
00:04:24,449 --> 00:04:27,928
approach to work. 
So taking this like very 

92
00:04:27,928 --> 00:04:31,883
disciplined focus approach to 
work and applying it in that, 

93
00:04:31,883 --> 00:04:34,541
and it mostly worked. 
And it actually had a lot of 

94
00:04:34,541 --> 00:04:36,409
benefits as well. 
So that was quite an experience 

95
00:04:36,409 --> 00:04:39,311
as well. 
Lastly, also related to my time 

96
00:04:39,311 --> 00:04:43,493
in Indonesia, I think was the 
BBM experience where the company

97
00:04:43,493 --> 00:04:46,661
I was working for back then 
Emtek decided to make a very big

98
00:04:46,661 --> 00:04:50,666
investment and buy what was 
remaining of the Blackberry 

99
00:04:50,666 --> 00:04:55,178
Messenger user base from 
Blackberry or RIM, but by that 

100
00:04:55,178 --> 00:04:57,101
point it had been renamed 
Blackberry. 

101
00:04:57,731 --> 00:05:02,122
And like relaunched it in 
Indonesia as like a local 

102
00:05:02,122 --> 00:05:03,971
messenger. 
It was very ambitious. 

103
00:05:04,422 --> 00:05:08,418
It was a crazy idea. 
It involved a lot of hard work 

104
00:05:08,418 --> 00:05:11,646
and it was a really, uh, 
eye-opening experience in terms 

105
00:05:11,646 --> 00:05:13,932
of kind of working, again, 
across cultures, like working 

106
00:05:13,932 --> 00:05:17,618
with a Canadian team, working 
across time zones. 

107
00:05:17,648 --> 00:05:19,680
And then there was a big 
technical challenge as well in 

108
00:05:19,680 --> 00:05:23,544
terms of doing a on-prem to 
cloud migration to Google Cloud 

109
00:05:23,544 --> 00:05:26,280
back then, which was also very 
new in the region. 

110
00:05:26,580 --> 00:05:29,099
So that was also quite an 
experience, kind of like taking 

111
00:05:29,099 --> 00:05:32,640
a bet on a new cloud provider, 
building relationships, 

112
00:05:32,640 --> 00:05:35,610
understand technical problems, 
solving tough technical 

113
00:05:35,610 --> 00:05:38,224
problems. 
Yeah, I think those three 

114
00:05:38,224 --> 00:05:39,823
things. 
And then I've had, I've been 

115
00:05:39,823 --> 00:05:42,432
very lucky, right? 
Like my move to Grab has been 

116
00:05:42,432 --> 00:05:44,015
quite exciting. 
I've been learning a lot as 

117
00:05:44,015 --> 00:05:46,600
well. 
But if I were to look back and 

118
00:05:46,600 --> 00:05:49,423
say like, okay, which were 
experiences that really shaped 

119
00:05:49,423 --> 00:05:52,672
who I am today from a career 
professional standpoint, it 

120
00:05:52,672 --> 00:05:54,397
would be those three aspects. 
Right. 

121
00:05:54,787 --> 00:05:56,347
Thank you so much for sharing 
this story. 

122
00:05:56,347 --> 00:06:00,890
I know you have had long, more 
eventful, you know, highlights 

123
00:06:00,890 --> 00:06:02,624
in your career, right? 
So I think thanks for 

124
00:06:02,624 --> 00:06:05,438
highlighting these three. 
I'm a bit intrigued by your 

125
00:06:05,438 --> 00:06:07,400
sharing about Pivotal Labs 
experience, right? 

126
00:06:07,790 --> 00:06:11,187
Because you were saying that you
were immersed in the practices 

127
00:06:11,187 --> 00:06:14,101
that they have, you know, XP, 
extreme programming, maybe a 

128
00:06:14,101 --> 00:06:17,455
little bit flavor of Agile, and 
more on the core technical 

129
00:06:17,455 --> 00:06:20,200
practices. 
These days, actually many people

130
00:06:20,200 --> 00:06:23,543
get exposed a lot more on the AI
tool stuff, right? 

131
00:06:23,543 --> 00:06:26,173
So not necessarily those kind of
technical practices. 

132
00:06:26,203 --> 00:06:29,150
In fact, it's more like a 
shortcut, like you can ask 

133
00:06:29,150 --> 00:06:31,531
something and it will give you 
the result, right? 

134
00:06:31,861 --> 00:06:35,771
So for people who are maybe 
starting in their journey, what 

135
00:06:35,771 --> 00:06:39,567
do you think still relevant? 
Looking back from your career 

136
00:06:39,567 --> 00:06:42,740
experience, right, when you 
studied first-hand those kind of

137
00:06:42,740 --> 00:06:45,256
practices. 
For those people who are more AI

138
00:06:45,256 --> 00:06:47,922
native, so to speak, right? 
Is there anything that you want 

139
00:06:47,922 --> 00:06:51,435
to convey to them such that they
can have a more better way of 

140
00:06:51,435 --> 00:06:52,366
learning? 
Yeah. 

141
00:06:52,726 --> 00:06:54,946
It's a really good question. 
It's definitely very topical. 

142
00:06:54,946 --> 00:06:58,522
Something that we discuss a lot 
at Grab, and with other friends 

143
00:06:58,522 --> 00:07:01,181
as well. 
You know, Kent Beck actually 

144
00:07:01,181 --> 00:07:04,459
said, and I'm paraphrasing here,
that after he saw like GenAI for

145
00:07:04,459 --> 00:07:08,386
the first time, he realized that
99% of what he knew became 

146
00:07:08,386 --> 00:07:10,735
irrelevant. 
But, or was it like 90% of what 

147
00:07:10,735 --> 00:07:13,823
he knew became irrelevant, but 
that 10% that he had learned 

148
00:07:13,823 --> 00:07:16,063
became a thousand times more 
important. 

149
00:07:16,093 --> 00:07:18,043
I'm paraphrasing here. 
Don't quote me on it. 

150
00:07:18,043 --> 00:07:21,283
But essentially what he was 
saying is that like a lot of 

151
00:07:21,283 --> 00:07:24,193
core skills are still extremely 
important, if not more important

152
00:07:24,193 --> 00:07:27,505
in this, in this realm. 
And I agree 100% there. 

153
00:07:27,535 --> 00:07:29,495
I think if you think about 
things like, especially test 

154
00:07:29,495 --> 00:07:31,790
driven development or pair 
programming. 

155
00:07:32,421 --> 00:07:34,870
And there's lots of material on 
this, I'm gonna repeat it. 

156
00:07:34,870 --> 00:07:37,400
But like, I think these are the 
type of techniques and 

157
00:07:37,400 --> 00:07:40,100
approaches that are going to 
help people use AI better. 

158
00:07:41,022 --> 00:07:44,652
I am not convinced that we're 
gonna go full autonomous just 

159
00:07:44,652 --> 00:07:46,900
yet. 
I think Andrej Karpathy talked 

160
00:07:46,900 --> 00:07:50,728
about it in his YC Combinator 
talk where he talked about this 

161
00:07:50,728 --> 00:07:55,663
sliding autonomy, I think we're 
very much still in a stage where

162
00:07:55,663 --> 00:08:00,217
AI can help us accelerate lots 
of work, help us with debugging,

163
00:08:00,217 --> 00:08:02,366
but the human's still gonna be 
in the loop. 

164
00:08:02,366 --> 00:08:06,224
So the question then is how do 
you make that loop something 

165
00:08:06,224 --> 00:08:09,548
where the human's providing 
guidance in a way that makes the

166
00:08:09,548 --> 00:08:12,993
AI do its job better. 
John Ousterhout actually talked 

167
00:08:12,993 --> 00:08:16,103
about this in the Pragmatic 
Engineer podcast as well, where 

168
00:08:16,103 --> 00:08:19,524
he talked about the value of 
good system design and system 

169
00:08:19,524 --> 00:08:23,017
architecture will probably get 
increased as we try to build 

170
00:08:23,017 --> 00:08:25,762
these AI systems. 
Because now the role of the 

171
00:08:25,762 --> 00:08:28,892
engineer is to kind of think 
about the architecture and the 

172
00:08:28,892 --> 00:08:30,918
design. 
Because the cost, the 

173
00:08:30,918 --> 00:08:33,770
incremental cost of producing 
code has, I mean, practically 

174
00:08:33,770 --> 00:08:35,918
gone to zero. 
The question now, are you 

175
00:08:35,918 --> 00:08:39,150
producing the right code? 
So I think a lot of these all, 

176
00:08:39,150 --> 00:08:42,337
like I wouldn't say it's all, 
but like a lot of these more 

177
00:08:42,337 --> 00:08:45,420
foundational skills around 
system design, test driven, uh, 

178
00:08:45,420 --> 00:08:48,530
like how do you apply tests to 
help with design? 

179
00:08:49,140 --> 00:08:53,859
How do you think about a 
feedback loop where you can 

180
00:08:53,859 --> 00:08:57,130
write some tests, write some 
code, refactor? 

181
00:08:57,570 --> 00:09:01,260
These are all very fundamental 
thinking that actually becomes 

182
00:09:01,260 --> 00:09:02,640
even more relevant in the AI 
world. 

183
00:09:02,670 --> 00:09:05,943
So I would definitely recommend 
folks who are looking at this 

184
00:09:05,943 --> 00:09:07,690
space. 
And yes, of course, everybody 

185
00:09:07,690 --> 00:09:09,430
enjoys their session of vibe 
coding. 

186
00:09:09,760 --> 00:09:13,326
And just kind of generating this
cool apps without giving it too 

187
00:09:13,326 --> 00:09:15,310
much more. 
And I think that's great. 

188
00:09:15,630 --> 00:09:17,954
Yeah. 
But I think I do agree, I think 

189
00:09:17,954 --> 00:09:19,560
Simon Williamson also talks 
about this. 

190
00:09:20,040 --> 00:09:21,330
That's a different type of 
coding. 

191
00:09:21,900 --> 00:09:24,213
The type of coding that you 
wanna do, that you wanna put in 

192
00:09:24,213 --> 00:09:26,070
production probably is not gonna
be vibe coding. 

193
00:09:26,670 --> 00:09:29,121
Right. 
So definitely, I mean from my 

194
00:09:29,121 --> 00:09:31,947
experience as well, I can see 
that the value of fundamentals 

195
00:09:31,947 --> 00:09:35,177
become much more important. 
Simply because now like what you

196
00:09:35,177 --> 00:09:37,207
said, producing code is so much 
easier. 

197
00:09:38,037 --> 00:09:39,337
Especially if you're a junior, 
right? 

198
00:09:39,387 --> 00:09:42,330
Even though you don't have much 
knowledge, you can still produce

199
00:09:42,330 --> 00:09:44,362
like so-called good code, I 
would say. 

200
00:09:44,839 --> 00:09:47,044
But whether the coherence in 
whether the terms of 

201
00:09:47,044 --> 00:09:48,676
architecture, design, 
performance, security aspects 

202
00:09:48,676 --> 00:09:51,889
and all that probably needs a 
little bit more fundamentals, 

203
00:09:51,889 --> 00:09:52,862
right? 
100%. 

204
00:09:53,108 --> 00:09:54,898
And you mentioned about pair 
programming. 

205
00:09:54,898 --> 00:09:58,997
Many people associate AI as like
my pair programmer now. 

206
00:09:59,187 --> 00:10:03,443
And they think even like pairing
with AI can give maybe the same 

207
00:10:03,443 --> 00:10:05,417
benefit. 
I know that the pair programming

208
00:10:05,417 --> 00:10:08,336
is not just having another 
person or thing working side by 

209
00:10:08,336 --> 00:10:10,973
side with you. 
I think it's the conversation, 

210
00:10:10,973 --> 00:10:14,474
the social aspect, maybe the two
perspectives talking to each 

211
00:10:14,474 --> 00:10:16,749
other and getting the best idea,
right? 

212
00:10:16,809 --> 00:10:21,129
So do you think pairing with AI 
can give you the same benefit? 

213
00:10:21,729 --> 00:10:23,895
So it depends. 
So if you think about pair 

214
00:10:23,895 --> 00:10:25,497
programming, there's several 
reasons why you do it. 

215
00:10:25,993 --> 00:10:30,371
If you go back to the original 
reason Pivotal Labs did it was, 

216
00:10:30,371 --> 00:10:34,225
uh, they wanted to all, they 
felt the best way to teach 

217
00:10:34,225 --> 00:10:36,753
someone was by teaching them on 
the job. 

218
00:10:36,843 --> 00:10:40,145
So what we would typically do is
when we had an engagement, we 

219
00:10:40,145 --> 00:10:44,284
would be pairing between a 
customer engineer and one the 

220
00:10:44,284 --> 00:10:49,479
Pivotal Labs consultant. 
I think that's that goal of 

221
00:10:49,479 --> 00:10:51,278
pairing. 
I mean, of course, when you're 

222
00:10:51,278 --> 00:10:52,904
talking about AI, you can't 
really teach AI. 

223
00:10:52,934 --> 00:10:55,163
We don't, we've not, we are not 
at a point where these models 

224
00:10:55,163 --> 00:10:56,564
can really take that feedback 
loop. 

225
00:10:56,774 --> 00:10:59,311
Yeah, you can write Cursor rules
and stuff like that, but it's 

226
00:10:59,311 --> 00:11:02,706
very different. 
I think that role of pairing is 

227
00:11:02,706 --> 00:11:05,259
a bit different. 
It's very relevant for 

228
00:11:05,259 --> 00:11:06,879
organizations that are trying to
upskill. 

229
00:11:06,909 --> 00:11:08,839
So like when we were in 
Indonesia and applying pair 

230
00:11:08,839 --> 00:11:11,121
programming, that was one of the
main reasons why we did it as 

231
00:11:11,121 --> 00:11:13,353
well. 
It was just we could onboard 

232
00:11:13,353 --> 00:11:15,722
someone onto a project within a 
week. 

233
00:11:16,479 --> 00:11:19,599
And they could be productive. 
Just accelerated learning and 

234
00:11:19,599 --> 00:11:20,865
exposure. 
Of course, there's cultural 

235
00:11:20,865 --> 00:11:22,875
aspects and stuff like that that
they need to gel with. 

236
00:11:23,295 --> 00:11:27,165
But if they did, you could take 
someone who's a fresh hire and 

237
00:11:27,165 --> 00:11:29,151
they could be committing code 
into production before the end 

238
00:11:29,151 --> 00:11:31,803
of the week. 
They could be writing a feature 

239
00:11:31,803 --> 00:11:33,917
and driving stuff by next week 
too. 

240
00:11:34,702 --> 00:11:36,262
Of course, you know, complexity 
and stuff like that. 

241
00:11:36,312 --> 00:11:38,237
So I think that's one aspect of 
pair programming. 

242
00:11:38,237 --> 00:11:41,748
Now, the I think the aspect of 
pair programming that is 

243
00:11:41,748 --> 00:11:45,876
probably a place where AI can 
play a role is around having 

244
00:11:45,876 --> 00:11:49,378
like a rubber duck where you 
have something that you can 

245
00:11:49,378 --> 00:11:51,776
throw ideas to and you can 
provide some feedback to the 

246
00:11:51,776 --> 00:11:53,542
ideas. 
So for example, like maybe you 

247
00:11:53,542 --> 00:11:56,062
have a design approach and you 
ask it like, hey, what about 

248
00:11:56,062 --> 00:11:59,296
this? 
The AI is pretty good at coming 

249
00:11:59,296 --> 00:12:01,294
back with suggestions. 
Now, not all the suggestions are

250
00:12:01,294 --> 00:12:03,549
gonna be great, so you're still 
gonna need to apply your 

251
00:12:03,549 --> 00:12:05,207
thinking. 
But even in the programming 

252
00:12:05,207 --> 00:12:07,687
world, you know, there's this 
concept called rubber duck, 

253
00:12:07,687 --> 00:12:10,183
rubber ducking, where you just 
have a rubber duck and you talk 

254
00:12:10,183 --> 00:12:11,726
to it. 
Of course, the AI is better than

255
00:12:11,726 --> 00:12:14,321
a rubber duck. 
So just process of being able to

256
00:12:14,321 --> 00:12:17,219
have something to bounce ideas 
off and get feedback on is very 

257
00:12:17,219 --> 00:12:19,519
powerful. 
And AI is a great way to do 

258
00:12:19,519 --> 00:12:22,800
that. 
Thirdly, I think AI as a, this 

259
00:12:22,800 --> 00:12:25,329
also happened in pair 
programming where sometimes we 

260
00:12:25,329 --> 00:12:29,303
work on a project where someone,
one of the pair has maybe more 

261
00:12:29,303 --> 00:12:31,607
domain knowledge, more technical
knowledge about the space. 

262
00:12:31,607 --> 00:12:34,477
So say for example, someone was 
working on a project and now we 

263
00:12:34,477 --> 00:12:38,305
need to like go and maybe build 
a feature that involves some 

264
00:12:38,305 --> 00:12:42,403
front end work. 
Now I might be more stronger on 

265
00:12:42,403 --> 00:12:45,244
my backend skills. 
I might not be so good on CSS or

266
00:12:45,244 --> 00:12:47,179
styling. 
The other person might have more

267
00:12:47,179 --> 00:12:49,801
skills on styling. 
So inevitably what you'd have is

268
00:12:49,801 --> 00:12:53,185
a situation where the, pairs are
supporting each other to kind of

269
00:12:53,185 --> 00:12:56,979
get the work, unit of work done.
But also one pair is driving 

270
00:12:56,979 --> 00:12:58,834
more when it comes to certain 
domains. 

271
00:12:59,299 --> 00:13:01,774
And this is where, I think, AI 
pair programming works really 

272
00:13:01,774 --> 00:13:03,681
well. 
With AI pair programming, what 

273
00:13:03,681 --> 00:13:07,493
you see is that, folks can now 
write code in languages that 

274
00:13:07,493 --> 00:13:09,243
maybe they're not super familiar
with. 

275
00:13:10,053 --> 00:13:12,186
That doesn't mean they're not 
good engineers, they're still 

276
00:13:12,186 --> 00:13:14,541
experienced. 
So they can look at a code, 

277
00:13:14,541 --> 00:13:17,235
evaluate it, understand whether 
this looks like good design, 

278
00:13:17,235 --> 00:13:19,677
good taste. 
But they might not know all the 

279
00:13:19,677 --> 00:13:21,165
syntax. 
They might not know, okay, this 

280
00:13:21,165 --> 00:13:22,680
is the Pythonic way of writing 
it. 

281
00:13:22,710 --> 00:13:25,418
Or in TypeScript, this is what 
you do when you try to build 

282
00:13:25,418 --> 00:13:27,216
this type of interface or 
structures. 

283
00:13:27,636 --> 00:13:29,466
That's why AI is also pretty 
useful. 

284
00:13:29,586 --> 00:13:32,896
So I think, yes, you're right. 
In some areas, AI can be very 

285
00:13:32,896 --> 00:13:36,024
helpful as pair programming. 
Some areas, it's serving a 

286
00:13:36,024 --> 00:13:37,421
different function. 
Right. 

287
00:13:38,131 --> 00:13:40,161
So definitely the last point 
that you mentioned, I 

288
00:13:40,161 --> 00:13:43,420
experienced it myself as well. 
So I used to come from like 

289
00:13:43,420 --> 00:13:45,562
heavy Java background. 
And then, you know, these days 

290
00:13:45,562 --> 00:13:48,620
there's so many exotic language,
all the youngsters want to use 

291
00:13:48,620 --> 00:13:51,575
those languages. 
So obviously, uh, there's a 

292
00:13:51,575 --> 00:13:54,982
little bit of learning curve to 
understand those, you know, more

293
00:13:54,982 --> 00:13:56,133
exotic programming languages, 
right. 

294
00:13:56,133 --> 00:13:58,233
So definitely using AI can help 
a lot. 

295
00:13:58,932 --> 00:14:02,149
So one thing that I foresee, I 
haven't done a lot of pair 

296
00:14:02,149 --> 00:14:04,815
programming these days 
especially as I move up into 

297
00:14:04,815 --> 00:14:09,533
more management role as well. 
I would say maybe the type of 

298
00:14:09,533 --> 00:14:13,956
pair programming people do now 
is not a pair but like three 

299
00:14:13,956 --> 00:14:16,497
things, you know. 
Three people and then with AI. 

300
00:14:17,167 --> 00:14:19,757
And I feel that there's a danger
sometimes maybe, even if you are

301
00:14:19,757 --> 00:14:22,507
pairing two persons you would 
still maybe outsource your 

302
00:14:22,507 --> 00:14:26,507
thinking a lot to the AI and you
kind of like, wrangle the code 

303
00:14:26,507 --> 00:14:29,470
that is generated by AI. 
Do you think there's a danger in

304
00:14:29,470 --> 00:14:31,384
doing this? 
Uh, and what should people do 

305
00:14:31,384 --> 00:14:33,507
better if there is? 
No, I-definitely think there's a

306
00:14:33,507 --> 00:14:36,522
danger in doing this. 
You need to be conscious that 

307
00:14:36,522 --> 00:14:40,851
the AI is just going to try as 
much as possible, at least the 

308
00:14:40,851 --> 00:14:42,131
way they're currently trained 
today. 

309
00:14:42,181 --> 00:14:46,202
And like, you know, I use Claude
4, I don't program as much as I 

310
00:14:46,202 --> 00:14:48,536
would like to. 
I don't program as much actively

311
00:14:48,536 --> 00:14:51,785
at work, but outside of it, I do
play around with this, with 

312
00:14:51,785 --> 00:14:53,956
Cursor. 
I enjoy using Cursor and Claude 

313
00:14:53,956 --> 00:14:56,830
4, and Google Gemini. 
Now the thing about these models

314
00:14:56,830 --> 00:14:59,446
is they love producing code. 
So they'll produce as much code 

315
00:14:59,446 --> 00:15:02,308
as they need to kind of answer 
the question and move on. 

316
00:15:03,064 --> 00:15:06,754
And the reality is, most of that
code sometimes without the right

317
00:15:06,754 --> 00:15:08,644
guardrails is gonna be 
gibberish. 

318
00:15:08,764 --> 00:15:11,199
Gibberish not in the sense that 
it doesn't work, but gibberish 

319
00:15:11,199 --> 00:15:13,834
in the sense that the next 
programmer who comes and looks 

320
00:15:13,834 --> 00:15:16,384
at it, they're gonna go, why are
we doing this? 

321
00:15:17,404 --> 00:15:20,965
Now again, it's a very common 
thing to say, but it's 

322
00:15:20,965 --> 00:15:24,826
definitely true, is that code is
read more by humans, is meant to

323
00:15:24,826 --> 00:15:28,080
be read by humans more than it 
is meant to be run by the 

324
00:15:28,080 --> 00:15:29,770
machine. 
Now, of course, you need it, you

325
00:15:29,770 --> 00:15:32,308
need the machine to run it. 
That's how you get business 

326
00:15:32,308 --> 00:15:35,446
value out of it. 
But the reason why we have so 

327
00:15:35,446 --> 00:15:38,421
many programming languages and 
different things is because we 

328
00:15:38,421 --> 00:15:41,864
are always looking for ways to 
express these concepts and ideas

329
00:15:41,864 --> 00:15:44,920
in a way that the machine can 
understand, which at the end of 

330
00:15:44,920 --> 00:15:48,161
the day is like, you know, you 
take it down to, you know, 

331
00:15:48,161 --> 00:15:52,263
microcode and you run it in the 
CPU after all the compilation or

332
00:15:52,263 --> 00:15:54,748
interpretative steps. 
But really what you're trying to

333
00:15:54,748 --> 00:15:56,710
design is for humans. 
And that aspect still applies. 

334
00:15:56,710 --> 00:16:01,796
And I'm not sure if the AI 
models today are trained enough 

335
00:16:01,796 --> 00:16:05,311
to have that point of view. 
They can definitely be guided to

336
00:16:05,311 --> 00:16:07,826
ensure that code is well written
and stuff like that. 

337
00:16:07,826 --> 00:16:10,766
But you need to play an active 
role in ensuring that happens. 

338
00:16:12,146 --> 00:16:15,568
So I think what I've seen and 
what I've heard and what 

339
00:16:15,568 --> 00:16:18,448
I've-tried that works is like, 
you know, the work needs to be 

340
00:16:18,448 --> 00:16:22,025
done very incrementally. 
Like this zero shot build an app

341
00:16:22,025 --> 00:16:25,300
thing, it can work. 
And I have to agree, it's 

342
00:16:25,300 --> 00:16:27,031
addictive. 
The dopamine hit of-being able 

343
00:16:27,031 --> 00:16:31,190
to go from idea to something 
that you can run in a Replit 

344
00:16:31,190 --> 00:16:33,661
window or deploy with Lovable is
amazing. 

345
00:16:34,292 --> 00:16:38,222
But yeah, I, I just don't know. 
I think it's good for zero shot.

346
00:16:38,222 --> 00:16:41,036
Like, oh, I need to build like a
quick guest registry for my 

347
00:16:41,036 --> 00:16:43,343
wedding. 
Or I want to, you know, build 

348
00:16:43,343 --> 00:16:45,107
something to keep track of 
bookmarks. 

349
00:16:45,197 --> 00:16:46,967
Yeah, for sure. 
Those throwaway apps. 

350
00:16:46,967 --> 00:16:50,491
But anything that you want to 
kind of build a business around 

351
00:16:50,491 --> 00:16:52,577
and maintain, you're going to 
need to be more involved. 

352
00:16:52,817 --> 00:16:54,455
Yeah. 
I think it's a very important 

353
00:16:54,455 --> 00:16:55,517
point that you mentioned, you 
know. 

354
00:16:55,517 --> 00:16:58,896
The code will still be read more
by humans even though produced 

355
00:16:58,896 --> 00:17:01,217
by AI or what, no matter what, 
right? 

356
00:17:01,547 --> 00:17:04,490
So in the end, someone needs to 
reason about the system, 

357
00:17:04,490 --> 00:17:06,457
especially for systems that 
evolve, right? 

358
00:17:06,457 --> 00:17:09,775
If like what you said, you just 
need to build it once, you know,

359
00:17:09,775 --> 00:17:11,957
like maybe wedding planner or 
whatever that is right? 

360
00:17:12,077 --> 00:17:14,717
Maybe you are not going to read 
it ever again. 

361
00:17:15,077 --> 00:17:18,057
But if, uh, we build software 
that is more like evolving, 

362
00:17:18,057 --> 00:17:20,894
serve a business purpose, I 
guess it will still be read a 

363
00:17:20,894 --> 00:17:22,823
lot by human. 
And we should not forget the 

364
00:17:22,823 --> 00:17:25,055
aspect that even though AI 
produce the code, someone needs 

365
00:17:25,055 --> 00:17:28,574
to understand what is being 
written at the end of the day. 

366
00:17:29,391 --> 00:17:32,375
The dopamine hit, I think, is 
really interesting you 

367
00:17:32,375 --> 00:17:34,947
mentioned. 
I personally find find it very 

368
00:17:34,947 --> 00:17:38,791
uh, addictive as well, right? 
Because maybe a long, long, time

369
00:17:38,791 --> 00:17:42,974
ago we used to think like, oh, I
want to be able to solve the 

370
00:17:42,974 --> 00:17:44,916
problem and write the algorithm 
myself, right? 

371
00:17:45,636 --> 00:17:48,319
But now it's more like the 
objective I think slightly 

372
00:17:48,319 --> 00:17:49,777
changed because of this dopamine
hit. 

373
00:17:49,997 --> 00:17:53,077
I wanna write a prompt such that
AI can produce the code that I 

374
00:17:53,077 --> 00:17:53,897
want. 
Yes. 

375
00:17:54,162 --> 00:17:55,662
I think this is also dangerous 
to me. 

376
00:17:55,782 --> 00:17:59,735
So we kind of like lose the 
aspect of, you know, actually 

377
00:17:59,735 --> 00:18:02,080
producing the code, the 
algorithm and the thinking 

378
00:18:02,080 --> 00:18:05,371
process behind it. 
But kind of like trying to find 

379
00:18:05,371 --> 00:18:07,909
the best prompt to get AI to 
produce it. 

380
00:18:08,504 --> 00:18:11,149
So again, like, I don't know 
whether you feel the same. 

381
00:18:11,149 --> 00:18:14,528
That's the first thing. 
And like what should we do in 

382
00:18:14,528 --> 00:18:17,587
order to restrain ourselves from
outsourcing a lot more of these 

383
00:18:17,587 --> 00:18:18,259
things? 
Yeah. 

384
00:18:18,259 --> 00:18:21,640
Uh, again, I, I, I don't think 
it's so much about one being 

385
00:18:21,640 --> 00:18:24,234
right or wrong. 
It's more like, okay, what's the

386
00:18:24,234 --> 00:18:25,977
purpose? 
If the purpose is to kind of 

387
00:18:25,977 --> 00:18:28,614
build a quick weekend app, then 
yeah, you should spend as much 

388
00:18:28,614 --> 00:18:31,825
time as you can to zero shot 
what you want out the door. 

389
00:18:31,975 --> 00:18:34,171
I think that's great. 
You know, it's... 

390
00:18:34,171 --> 00:18:37,675
But if what you're trying to do 
is build something for your 

391
00:18:37,675 --> 00:18:40,975
professional work or you, like, 
you said, you wanna build a 

392
00:18:40,975 --> 00:18:43,729
system that's gonna continue to 
need to evolve, right? 

393
00:18:43,729 --> 00:18:48,396
It's not just a zero shot system
that kind of is deployed and 

394
00:18:48,396 --> 00:18:52,377
done, then there are lots of 
folks who have shared 

395
00:18:52,377 --> 00:18:54,152
approaches. 
I think the one that's worked 

396
00:18:54,152 --> 00:18:57,441
for me, and I think based on 
what I've read also works for a 

397
00:18:57,441 --> 00:18:59,489
lot of other people, is being 
very deliberate. 

398
00:18:59,549 --> 00:19:02,591
Like so spending time to first 
like kind of lay out a bit of an

399
00:19:02,591 --> 00:19:05,534
architecture or a design. 
Of course, at that point not 

400
00:19:05,534 --> 00:19:08,117
producing any code, just kind of
laying out what the design and 

401
00:19:08,117 --> 00:19:10,208
architecture should look like, 
how you think about things. 

402
00:19:10,598 --> 00:19:14,288
And then only producing code in 
very incremental fashion. 

403
00:19:14,288 --> 00:19:16,398
So saying like, okay, now that 
we've figured out this is what 

404
00:19:16,398 --> 00:19:18,576
we're gonna do, let's start with
this piece here. 

405
00:19:18,606 --> 00:19:21,846
I wanna build this screen and I 
want a controller that'll be 

406
00:19:21,846 --> 00:19:25,386
taking these inputs and let's 
also have a test for it, right? 

407
00:19:25,386 --> 00:19:29,300
I think doing those steps in 
increments is the right way to 

408
00:19:29,300 --> 00:19:32,935
do it. 
I've heard of some people who 

409
00:19:32,935 --> 00:19:36,657
have been more successful with 
like a TDD approach with AI. 

410
00:19:36,657 --> 00:19:41,007
So getting AI to write the test 
first, I'm not as sure. 

411
00:19:41,007 --> 00:19:43,107
I mean, my personal experience 
is it hasn't. 

412
00:19:43,257 --> 00:19:46,959
I haven't been able to be as 
successful in it as much as I 

413
00:19:46,959 --> 00:19:49,422
enjoy TDD. 
I actually find it more 

414
00:19:49,422 --> 00:19:51,822
practical for me maybe to write 
the test. 

415
00:19:52,679 --> 00:19:55,166
So maybe that's just more on how
we think or push things. 

416
00:19:55,166 --> 00:19:57,790
But definitely, test writing is 
an area where AI is very, very 

417
00:19:57,790 --> 00:20:00,218
good at. 
And like, you know, it, it'll 

418
00:20:00,218 --> 00:20:02,636
help with a lot of the, 
sometimes, the more tedious 

419
00:20:02,636 --> 00:20:05,514
aspects of like, you know, 
keeping the mocks up to date, 

420
00:20:05,514 --> 00:20:07,514
updating stuff. 
So like things that you would 

421
00:20:07,514 --> 00:20:10,066
normally write macros to go in, 
like quickly do refactors now, 

422
00:20:10,066 --> 00:20:12,266
now you can just get the LLM to 
do it. 

423
00:20:12,626 --> 00:20:14,840
So yeah, I think more 
incremental, more deliberate 

424
00:20:14,840 --> 00:20:17,186
thinking. 
Definitely, for the second 

425
00:20:17,186 --> 00:20:19,246
category of work. 
Yeah. 

426
00:20:19,786 --> 00:20:22,481
And not to mention, these days 
people are building more agentic

427
00:20:22,481 --> 00:20:26,085
way of coding and even like 
submitting a task asynchronously

428
00:20:26,085 --> 00:20:29,881
and then they'll just create a 
pull request that we just merge.

429
00:20:30,331 --> 00:20:33,601
I think the advancement here is 
really rapid, I would say. 

430
00:20:33,601 --> 00:20:36,041
Sometimes I also fear that I'm 
obsolete. 

431
00:20:36,866 --> 00:20:40,115
But yeah, I think let's move on 
to the next segue, because I 

432
00:20:40,115 --> 00:20:43,061
know that you have been in 
multiple different companies and

433
00:20:43,061 --> 00:20:46,096
you're kind of like considered 
very successful and people find 

434
00:20:46,096 --> 00:20:48,071
you a very inspiring tech 
leader, right? 

435
00:20:48,431 --> 00:20:52,535
So one aspect that I would want 
to for us to learn is how to 

436
00:20:52,535 --> 00:20:55,321
build that transformational 
engineering teams, right? 

437
00:20:55,641 --> 00:20:58,651
So maybe if you can, I don't 
know, summarize a few key 

438
00:20:58,651 --> 00:21:01,565
points. 
How do you actually think about 

439
00:21:01,565 --> 00:21:03,781
making a successful 
transformational engineering 

440
00:21:03,781 --> 00:21:04,871
team? 
That would be great. 

441
00:21:05,581 --> 00:21:07,300
Yeah. 
So first off, you're being very 

442
00:21:07,300 --> 00:21:08,454
generous. 
Thank you, Henry. 

443
00:21:08,797 --> 00:21:12,363
I think what I have been, over 
my career, is I've been very 

444
00:21:12,363 --> 00:21:15,367
lucky to work with very capable 
people. 

445
00:21:15,727 --> 00:21:19,791
So that has definitely helped. 
Having said that, I do have a 

446
00:21:19,791 --> 00:21:21,524
point of view on what makes good
teams. 

447
00:21:21,584 --> 00:21:23,810
Of course, it's based on my 
experience and experience will 

448
00:21:23,810 --> 00:21:25,094
vary, right? 
People have different approaches

449
00:21:25,094 --> 00:21:28,463
for these sort of things. 
And it's probably also because 

450
00:21:28,463 --> 00:21:32,026
of the Pivotal Labs, extreme 
programming type thinking. 

451
00:21:32,116 --> 00:21:36,706
But I feel like high performing 
engineering teams are very 

452
00:21:36,706 --> 00:21:38,969
similar to high performing 
sports teams. 

453
00:21:39,639 --> 00:21:41,973
What do I mean by that? 
Firstly, especially when you're 

454
00:21:41,973 --> 00:21:44,484
talking about like complex 
engineering problems, which 

455
00:21:44,484 --> 00:21:48,573
can't just be solved by one, two
person teams, it's a team sport.

456
00:21:48,933 --> 00:21:50,913
You need all the different 
players. 

457
00:21:50,973 --> 00:21:54,358
They're bringing different 
skills or capabilities to the 

458
00:21:54,358 --> 00:21:56,315
field. 
Similarly in an engineering 

459
00:21:56,315 --> 00:21:58,447
project, there's gonna be the 
person who goes deep. 

460
00:21:58,477 --> 00:22:02,161
There's gonna be the person who 
is very good at kind of 

461
00:22:02,161 --> 00:22:03,997
communicating and being the 
bridge between teams. 

462
00:22:04,477 --> 00:22:07,717
Uh, you know, there's gonna be 
someone who's able to rally 

463
00:22:07,717 --> 00:22:11,353
people when things are down. 
So you need all sorts of 

464
00:22:11,353 --> 00:22:14,299
characters in the group. 
A diverse set of people is 

465
00:22:14,299 --> 00:22:16,319
important. 
On the back of that, of course, 

466
00:22:16,319 --> 00:22:18,438
you have things like frank and 
open communication, discipline, 

467
00:22:18,438 --> 00:22:23,581
respect, and all these sort of 
aspects as well, which I guess 

468
00:22:23,581 --> 00:22:25,760
is common for most high 
performing teams, whether 

469
00:22:25,760 --> 00:22:27,712
they're software engineering or 
not, right? 

470
00:22:28,732 --> 00:22:32,839
But what I think is maybe a bit 
more distinctive is all of this 

471
00:22:32,839 --> 00:22:36,992
needs to be supported by a 
foundation of like some level of

472
00:22:36,992 --> 00:22:40,025
discipline, which is not 
something we normally associate 

473
00:22:40,025 --> 00:22:43,932
with our line of work. 
I think highly effective teams 

474
00:22:43,932 --> 00:22:45,510
are disciplined. 
Now that doesn't mean everybody 

475
00:22:45,510 --> 00:22:47,902
comes in at seven o'clock in the
morning and all that kind of 

476
00:22:47,902 --> 00:22:49,880
stuff, although that's what they
did do in Pivotal, there was 

477
00:22:49,880 --> 00:22:53,254
like fixed timings of the day. 
It does mean though, there are 

478
00:22:53,254 --> 00:22:55,742
certain bars of expectation that
everybody adheres to. 

479
00:22:55,772 --> 00:22:59,237
So maybe things like, okay, the,
our code quality, our testing 

480
00:22:59,237 --> 00:23:02,342
approach, uh, how we respond to 
incidents. 

481
00:23:03,117 --> 00:23:06,861
That's what I think sets the 
best teams apart. 

482
00:23:08,091 --> 00:23:10,869
There are other aspects, so 
like, so when you have like a 

483
00:23:10,869 --> 00:23:15,278
high regard for like engineering
quality, you naturally are also 

484
00:23:15,278 --> 00:23:17,391
aspiring each other to push your
limits. 

485
00:23:17,391 --> 00:23:20,057
That's something I felt is very 
important. 

486
00:23:20,267 --> 00:23:24,625
So when you have like strong 
engineers on the team, it lifts 

487
00:23:24,625 --> 00:23:27,722
everybody else as well. 
And it creates a feedback loop. 

488
00:23:27,812 --> 00:23:30,984
So now everybody gets better. 
So now people push harder to, to

489
00:23:30,984 --> 00:23:33,921
learn even more, to solve 
problems in an even better way. 

490
00:23:34,311 --> 00:23:37,061
And that feedback loop of 
learning from each other, 

491
00:23:37,061 --> 00:23:39,381
teaching each other, that's 
really important. 

492
00:23:40,231 --> 00:23:44,261
So another thing that I think is
really important is having a 

493
00:23:44,261 --> 00:23:47,441
good mix of juniors, medium 
folks, and senior folks on 

494
00:23:47,441 --> 00:23:49,340
teams. 
I know there's a lot more focus 

495
00:23:49,340 --> 00:23:51,733
now, like, oh, are we getting 
rid of juniors and all that? 

496
00:23:51,733 --> 00:23:53,783
Like I don't fully buy into 
that. 

497
00:23:53,783 --> 00:23:57,105
I just don't think that's a 
sustainable way to set an 

498
00:23:57,105 --> 00:23:59,742
engineering team up. 
You can maybe do that for like 

499
00:23:59,742 --> 00:24:00,471
very small teams. 
Okay. 

500
00:24:00,471 --> 00:24:04,238
Like I've got a very small team 
of just like, super elite 

501
00:24:04,238 --> 00:24:05,782
engineers. 
Sure. 

502
00:24:05,972 --> 00:24:08,299
But like if you're talking about
like, you know, 30, 40, 50 

503
00:24:08,299 --> 00:24:10,922
person org, you're gonna have 
varying levels of skill and 

504
00:24:10,922 --> 00:24:12,282
experience. 
And people are going to kind of 

505
00:24:12,282 --> 00:24:15,028
go up the chain. 
I think what we mean when we say

506
00:24:15,028 --> 00:24:17,740
stuff like, we are not gonna 
hire junior engineers anymore, 

507
00:24:17,740 --> 00:24:22,363
is that we are raising the bar 
of what a junior engineer should

508
00:24:22,363 --> 00:24:25,371
be capable of. 
But when you raise the bar at 

509
00:24:25,371 --> 00:24:26,939
the bottom, of course, the top 
moves up. 

510
00:24:27,329 --> 00:24:31,919
So the relative skills, I think,
gap between these different 

511
00:24:31,919 --> 00:24:35,573
levels will still be there. 
And that in that you'll create 

512
00:24:35,573 --> 00:24:38,600
an opportunity for good team 
dynamics of mentorship, 

513
00:24:38,600 --> 00:24:42,789
coaching, and learning, which I 
think is very critical in good 

514
00:24:42,789 --> 00:24:44,228
teams. 
And that's what I've seen, 

515
00:24:44,228 --> 00:24:46,029
right? 
Like, my experience is that 

516
00:24:46,029 --> 00:24:49,733
people who join and stay with 
teams, especially at a younger 

517
00:24:49,733 --> 00:24:52,218
age. 
The more junior folks who come 

518
00:24:52,218 --> 00:24:53,640
in, they, they're just more 
hungry. 

519
00:24:54,390 --> 00:24:56,670
They're willing to do anything 
to kind of learn. 

520
00:24:56,670 --> 00:24:58,724
Especially, if there's an 
opportunity to learn from 

521
00:24:58,724 --> 00:25:00,870
someone who's a mentor or a 
coach. 

522
00:25:01,110 --> 00:25:03,420
And then the more senior folks, 
maybe they're a bit more 

523
00:25:03,420 --> 00:25:05,340
specialized, so there's certain 
areas that they want to kind of 

524
00:25:05,340 --> 00:25:07,608
go deeper on. 
But they're more willing to 

525
00:25:07,608 --> 00:25:10,479
teach as well and to get a 
reward from doing that. 

526
00:25:10,479 --> 00:25:14,993
So yeah, those are some aspects 
I feel of like highly performing

527
00:25:14,993 --> 00:25:15,819
teams. 
Yeah. 

528
00:25:15,969 --> 00:25:18,942
And not to mention, like many 
people associate also software 

529
00:25:18,942 --> 00:25:21,312
engineering, like a 
apprenticeship model, right? 

530
00:25:21,312 --> 00:25:25,240
Where you have someone senior 
coaching you, bringing along, 

531
00:25:25,240 --> 00:25:27,966
solving problems together until 
a certain point where they 

532
00:25:27,966 --> 00:25:31,182
become, I dunno, much more 
senior and then they can go on 

533
00:25:31,182 --> 00:25:32,802
by themselves as a leader, 
right? 

534
00:25:33,332 --> 00:25:36,346
So interestingly, you mentioned 
about discipline and associate 

535
00:25:36,346 --> 00:25:40,094
that with like sports team, I 
think in a physical sports team 

536
00:25:40,094 --> 00:25:43,044
is very clear. 
Discipline is about, you know, 

537
00:25:43,044 --> 00:25:46,154
time, physical conditions. 
You can even measure and look at

538
00:25:46,154 --> 00:25:49,017
it very concretely, right? 
Whether you're fit, not fit, fat

539
00:25:49,017 --> 00:25:51,632
or you know, like muscular. 
I think it's very easy. 

540
00:25:51,632 --> 00:25:54,072
But for engineering team, 
probably it's a little bit more 

541
00:25:54,072 --> 00:25:58,337
abstract and very hard to see. 
And in fact, many people try now

542
00:25:58,337 --> 00:26:00,356
to also measure the 
effectiveness of their 

543
00:26:00,356 --> 00:26:02,222
engineering team, the 
productivity and all that. 

544
00:26:02,672 --> 00:26:05,192
Sometimes it's arguable this 
measure against the others. 

545
00:26:05,342 --> 00:26:08,428
But in your point of view, if 
you can distill this discipline,

546
00:26:08,428 --> 00:26:11,902
right, what kind of things that 
you would measure or would you 

547
00:26:11,902 --> 00:26:15,587
set as the high bar? 
Yeah, so some aspects of it is 

548
00:26:15,587 --> 00:26:17,622
engineering culture, like what 
we talked about, right? 

549
00:26:17,622 --> 00:26:21,452
So like things like, okay, how 
do you... like code quality, how

550
00:26:21,452 --> 00:26:24,582
you're, uh, which is both like 
in terms of like readability of 

551
00:26:24,582 --> 00:26:27,092
your code, design of your code, 
testing. 

552
00:26:27,529 --> 00:26:30,079
Also like, just, yeah, your 
entire deployment process. 

553
00:26:30,079 --> 00:26:33,984
Like how do you, like how 
effective are you around 

554
00:26:33,984 --> 00:26:36,354
deployments? 
How does team responds to 

555
00:26:36,354 --> 00:26:38,334
incidents or issues in 
production? 

556
00:26:38,724 --> 00:26:40,674
What's the culture? 
Is it a blameless culture? 

557
00:26:40,674 --> 00:26:42,924
Is it a culture of learning, 
right? 

558
00:26:43,224 --> 00:26:47,100
I think another aspect that I 
feel really set teams apart, and

559
00:26:47,100 --> 00:26:51,046
something that we try to instill
early on when we, when I've run 

560
00:26:51,046 --> 00:26:55,253
teams together with other 
leaders is that you must instill

561
00:26:55,253 --> 00:27:00,683
a culture of wanting to get into
the technical details. 

562
00:27:01,073 --> 00:27:05,441
Like no issue is too deep. 
So, you know, in Indonesia, we 

563
00:27:05,441 --> 00:27:07,458
used to like to say, uh, no 
magic. 

564
00:27:08,353 --> 00:27:10,843
So what that means is like let's
say there's a bug in production.

565
00:27:11,453 --> 00:27:14,613
And it's flaky, and it sometimes
happens, it doesn't happen. 

566
00:27:15,323 --> 00:27:18,201
When we took over the team, 
there was some older engineers 

567
00:27:18,201 --> 00:27:20,233
around, they'll say like, oh, I 
don't know, black magic. 

568
00:27:20,813 --> 00:27:22,353
And so we, we just were like, 
No! 

569
00:27:22,413 --> 00:27:24,918
No magic! 
Let's find out what's wrong 

570
00:27:24,918 --> 00:27:27,271
here. 
Let's put in the right level of 

571
00:27:27,271 --> 00:27:29,258
debugging logs. 
Let's make sure there's enough 

572
00:27:29,258 --> 00:27:31,128
observability. 
Let's try to reproduce this 

573
00:27:31,128 --> 00:27:32,943
locally. 
Let's try to like, okay, no, 

574
00:27:32,943 --> 00:27:35,403
this is in the C code. 
All right, let's read through 

575
00:27:35,403 --> 00:27:37,664
the C code. 
Let's get the open source 

576
00:27:37,664 --> 00:27:40,337
version of this application. 
Let's look through the code. 

577
00:27:40,667 --> 00:27:43,918
Let's see if we can modify it. 
Oh, you mean it's a language of 

578
00:27:43,918 --> 00:27:45,527
framework you've not used? 
Okay, you got two weeks. 

579
00:27:45,527 --> 00:27:47,447
Go figure it out. 
Like, go deep. 

580
00:27:47,747 --> 00:27:49,967
Don't stop at just the surface 
level. 

581
00:27:50,387 --> 00:27:53,947
And, you know, sometimes it'll 
feel like, wow, so much effort. 

582
00:27:54,097 --> 00:27:55,417
It's like this doesn't really 
affect. 

583
00:27:55,507 --> 00:27:58,399
But like I think that's what 
differentiates the teams that 

584
00:27:58,399 --> 00:28:02,936
have the right type of rigor and
discipline around focusing on 

585
00:28:02,936 --> 00:28:05,738
their craft and really 
understanding how things work. 

586
00:28:06,098 --> 00:28:07,778
Another thing is like investment
in tools. 

587
00:28:07,808 --> 00:28:10,466
I think that's another area. 
The discipline to kind of go 

588
00:28:10,466 --> 00:28:13,491
like, hey, we're not just gonna 
jump from one tool to the other 

589
00:28:13,491 --> 00:28:15,424
tool. 
Like we have a tool, we're gonna

590
00:28:15,424 --> 00:28:17,168
go deep on it, we're gonna 
maximize it. 

591
00:28:17,558 --> 00:28:20,715
And then if we figure something 
else comes along, if it's really

592
00:28:20,715 --> 00:28:23,046
that much better, then we'll 
kind of be more deliberate about

593
00:28:23,046 --> 00:28:24,849
it, right? 
So like getting really good at 

594
00:28:24,849 --> 00:28:27,153
your tools, that's something 
Pivotal did really well as well.

595
00:28:27,693 --> 00:28:31,109
They just super invested in 
Rails and Ruby back then. but 

596
00:28:31,109 --> 00:28:32,655
they got so much mileage out of 
it. 

597
00:28:32,655 --> 00:28:35,391
Just cause they were super 
experts and they knew how to 

598
00:28:35,391 --> 00:28:38,365
make it do anything they needed 
to do and they knew the best way

599
00:28:38,365 --> 00:28:41,258
to kind of build it. 
So yeah, I think those are some 

600
00:28:41,258 --> 00:28:43,404
aspects that I would say 
separate those teams. 

601
00:28:43,584 --> 00:28:45,486
Yeah. 
So definitely culture and still 

602
00:28:45,486 --> 00:28:49,200
kind of like the code quality 
aspect, the deployment aspect, 

603
00:28:49,200 --> 00:28:52,720
the incident management aspect. 
The blamelessness thing, 

604
00:28:52,720 --> 00:28:55,224
obviously, right. 
So I think that goes back to the

605
00:28:55,224 --> 00:28:57,648
culture. 
And I like the no black magic 

606
00:28:57,648 --> 00:28:59,766
thing, right? 
Because sometimes, you know, 

607
00:28:59,766 --> 00:29:03,534
software and distributed systems
is very hard to reason about 

608
00:29:03,534 --> 00:29:05,632
sometimes. 
But yeah, I mean at the end of 

609
00:29:05,632 --> 00:29:08,541
the day I think we can always 
find an explanation even though 

610
00:29:08,541 --> 00:29:11,259
sometimes it's maybe, I dunno, 
concurrency or some random 

611
00:29:11,259 --> 00:29:13,964
things could happen. 
So I like that approach that you

612
00:29:13,964 --> 00:29:16,070
mentioned. 
The other aspect of engineering 

613
00:29:16,070 --> 00:29:18,660
team obviously is to deliver 
results for the business, right?

614
00:29:18,660 --> 00:29:21,988
And I think, I know that you 
have a few thoughts about this, 

615
00:29:21,988 --> 00:29:24,390
like always understanding about 
outcomes, small team. 

616
00:29:24,390 --> 00:29:27,255
Maybe a little bit on these 
aspects as well when you build 

617
00:29:27,255 --> 00:29:29,692
engineering team. 
Yeah, this is definitely 

618
00:29:29,692 --> 00:29:33,606
something that has become very 
much more focused for me now in 

619
00:29:33,606 --> 00:29:36,278
my time at Grab. 
Cause I run the platform 

620
00:29:36,278 --> 00:29:38,864
engineering teams here. 
And within our platform 

621
00:29:38,864 --> 00:29:41,524
engineering teams, we don't 
really have too many product 

622
00:29:41,524 --> 00:29:43,472
managers. 
We have and they're helpful and 

623
00:29:43,472 --> 00:29:46,178
they're good and it's a growing 
group and we enjoy working with 

624
00:29:46,178 --> 00:29:49,460
them. 
But what we typically find is 

625
00:29:49,460 --> 00:29:53,776
the engineers need to be the 
ones also kind of thinking about

626
00:29:53,776 --> 00:29:56,063
the customer. 
And the customer here being 

627
00:29:56,063 --> 00:29:59,958
internal service teams. 
Now, in a way that should be 

628
00:29:59,958 --> 00:30:02,683
easier because you should be 
able to understand that persona,

629
00:30:02,683 --> 00:30:06,515
cause you are that persona too. 
But it's interesting how hard it

630
00:30:06,515 --> 00:30:10,178
is sometimes for engineers to 
kind put themself in the shoe of

631
00:30:10,178 --> 00:30:13,166
the customer. 
I think there's a tendency to 

632
00:30:13,166 --> 00:30:14,708
get very fixated on the 
solution. 

633
00:30:14,858 --> 00:30:17,916
Like oh this is a cool solution 
or we can build it out this way 

634
00:30:17,916 --> 00:30:21,002
build out that way. 
And you end up in this trap of 

635
00:30:21,002 --> 00:30:23,798
just building widgets. 
So you got all these different 

636
00:30:23,798 --> 00:30:26,678
widgets and like, look at all 
this cool stuff we've built. 

637
00:30:27,098 --> 00:30:30,698
Now is that really solving the 
problem that we set out to 

638
00:30:30,698 --> 00:30:33,183
build, that you want to solve in
the first place? 

639
00:30:33,968 --> 00:30:38,163
And many a times when you ask 
the engineer that, you know, 

640
00:30:38,163 --> 00:30:42,158
they go like, oh, I haven't 
really given enough thought. 

641
00:30:42,158 --> 00:30:45,068
Or they might anchor one thing, 
oh, this so and so told me this.

642
00:30:45,578 --> 00:30:49,278
That's why I'm building this. 
And all well-intentioned, but 

643
00:30:49,278 --> 00:30:54,661
this is where I think, you know,
really focusing more on the 

644
00:30:54,661 --> 00:30:56,444
outcome. 
Like what is the problem we're 

645
00:30:56,444 --> 00:30:59,118
trying to solve? 
What are the different ways to 

646
00:30:59,118 --> 00:31:01,790
solve it? 
And then like, you know, from an

647
00:31:01,790 --> 00:31:03,732
effort standpoint, how much 
effort we're gonna put into 

648
00:31:03,732 --> 00:31:05,868
this? 
And okay, will this, and then 

649
00:31:05,868 --> 00:31:09,627
circling back again, right? 
So, you know, sometimes people 

650
00:31:09,627 --> 00:31:12,859
say product thinking. 
So I think especially in this AI

651
00:31:12,859 --> 00:31:16,081
world, the role of the software 
engineer and the product manager

652
00:31:16,081 --> 00:31:17,971
to some degree is melding, 
right? 

653
00:31:17,971 --> 00:31:21,844
Like, uh, I think there's this 
phrase called tiny teams is 

654
00:31:21,844 --> 00:31:25,385
becoming more popular. 
The Latent Space guys kind of 

655
00:31:25,385 --> 00:31:28,043
popularized it with their AI 
welfare. 

656
00:31:28,103 --> 00:31:29,783
They had a dedicated track on 
tiny teams. 

657
00:31:31,043 --> 00:31:34,518
And yeah, I think what that 
means is now that you have a 

658
00:31:34,518 --> 00:31:37,433
system that can help you produce
a code, you should be thinking 

659
00:31:37,433 --> 00:31:39,759
more about the design, more 
about the customer problem. 

660
00:31:40,489 --> 00:31:44,270
And it's definitely a muscle. 
We would love to have more 

661
00:31:44,270 --> 00:31:45,566
engineers build. 
Yeah. 

662
00:31:46,003 --> 00:31:50,512
I can actually see as well, if 
the engineers having that this 

663
00:31:50,512 --> 00:31:53,812
business mindset, wanting to 
understand the actual problem 

664
00:31:53,812 --> 00:31:56,890
user is facing, it's like a 
multiplier effect, right? 

665
00:31:56,890 --> 00:31:59,500
They understand the technicals, 
but they also understand like 

666
00:31:59,500 --> 00:32:02,470
the actual business outcomes 
that needs to be produced. 

667
00:32:03,116 --> 00:32:05,852
And especially these days, you 
mentioned right, tiny teams, you

668
00:32:05,852 --> 00:32:09,415
can kind of like outsource a lot
of mundane work to something 

669
00:32:09,415 --> 00:32:12,569
else like AI, right? 
I think it becomes more crucial 

670
00:32:12,569 --> 00:32:15,686
that people don't just focus on 
the cool stuff, but actually 

671
00:32:15,686 --> 00:32:19,331
solving the problem, getting the
right outcome for the team. 

672
00:32:19,631 --> 00:32:24,046
I think that's very important. 
And some of these is the kind of

673
00:32:24,046 --> 00:32:27,073
like the responsibility for 
leaders to kind of like guide 

674
00:32:27,073 --> 00:32:29,117
them, right? 
And I know uh, as a leader as 

675
00:32:29,117 --> 00:32:31,081
well, you have a lot of 
experience in this area. 

676
00:32:31,681 --> 00:32:36,274
So interestingly enough, um, you
know, like when people say about

677
00:32:36,274 --> 00:32:38,847
you inspiring leader, 
transformational, a lot of 

678
00:32:38,847 --> 00:32:42,109
things, they mentioned about the
tough approach that you have as 

679
00:32:42,109 --> 00:32:45,055
well, right? 
So tell us a little bit more 

680
00:32:45,055 --> 00:32:47,055
about your personal style as a 
leader, right? 

681
00:32:47,055 --> 00:32:50,985
How do you actually set the 
standards very high to the team?

682
00:32:50,985 --> 00:32:54,051
How do you communicate that? 
How do you actually train them 

683
00:32:54,051 --> 00:32:56,415
or mentor them to become the 
best software engineering team 

684
00:32:56,415 --> 00:33:00,345
that they could be? 
Yeah, it's a very good and 

685
00:33:00,345 --> 00:33:03,710
relevant question as well. 
You know, we're going through 

686
00:33:03,710 --> 00:33:06,202
our review period, so I'm sure 
I'm gonna see a lot of feedback 

687
00:33:06,202 --> 00:33:09,842
on stuff like this as well. 
Listen, I think, to be 

688
00:33:09,842 --> 00:33:13,367
completely honest, I think I'm 
also kind of calibrating and 

689
00:33:13,367 --> 00:33:17,617
figuring out what works best. 
I think a lot of like this 

690
00:33:17,617 --> 00:33:21,947
notion of like being a strong, a
tough leader, a lot of it, many 

691
00:33:21,947 --> 00:33:24,806
times comes from also like your 
experience learning. 

692
00:33:25,010 --> 00:33:29,090
So like when I was in school and
when I was in the debate team 

693
00:33:29,090 --> 00:33:32,205
and when, and also like, you 
know, subsequent to that, my 

694
00:33:32,205 --> 00:33:35,362
experience has always been 
around coaches that I kind of 

695
00:33:35,362 --> 00:33:37,447
like sports coaches. 
Like you, you know, you think 

696
00:33:37,447 --> 00:33:39,565
about sports coaches, they're 
also kind of like very strict. 

697
00:33:39,565 --> 00:33:41,545
Like, you know, because it comes
with a discipline regime. 

698
00:33:41,825 --> 00:33:45,025
That's kind of the type of 
teachers I've had my entire 

699
00:33:45,025 --> 00:33:47,374
life. 
My most effective teachers were 

700
00:33:47,374 --> 00:33:49,800
strong. 
Very clear on like, okay, this 

701
00:33:49,800 --> 00:33:52,644
is what we need to get done. 
This-is-how we're gonna do it. 

702
00:33:52,906 --> 00:33:56,280
And then if you don't do it, 
it's not so much like admonish 

703
00:33:56,280 --> 00:33:59,236
you, but like saying like, no, 
you can do better, right? 

704
00:33:59,266 --> 00:34:02,482
And I've, I guess, you know, 
very consciously, unconsciously,

705
00:34:02,482 --> 00:34:07,111
I've kind of modeled some of my 
approach to that as well to 

706
00:34:07,111 --> 00:34:09,751
varying different levels of 
success, to be completely 

707
00:34:09,751 --> 00:34:12,181
honest. 
I think when you have this type 

708
00:34:12,181 --> 00:34:15,540
of approach, you tend to attract
a certain class of folks to join

709
00:34:15,540 --> 00:34:18,445
your team as well, because 
people who don't kind of gel 

710
00:34:18,445 --> 00:34:19,945
with this style, they'll, 
they'll leave. 

711
00:34:20,340 --> 00:34:24,547
They'll move on. 
Now, the good thing though is 

712
00:34:24,547 --> 00:34:28,926
coincidentally, it has been a 
good filter, I would say. 

713
00:34:29,255 --> 00:34:32,746
I've been able to retain strong 
people as a result of that 

714
00:34:32,746 --> 00:34:34,536
because they also are driven by 
this thing. 

715
00:34:34,536 --> 00:34:37,176
They also want the clarity. 
They want the strong execution. 

716
00:34:37,176 --> 00:34:38,346
They're willing to put in the 
work. 

717
00:34:39,231 --> 00:34:41,937
I think though, where it might 
be like a bit of not as 

718
00:34:41,937 --> 00:34:45,771
effective, it's like, okay, can 
I help people be more effective?

719
00:34:45,891 --> 00:34:48,783
Or is this filter going to mean 
that you either fit into the 

720
00:34:48,783 --> 00:34:51,289
mold or you don't? 
And that's where I think I'm 

721
00:34:51,289 --> 00:34:52,641
trying to calibrate as a leader 
as well. 

722
00:34:53,031 --> 00:34:56,254
Like, okay, if you don't 
normally gel with this style, 

723
00:34:56,254 --> 00:35:00,150
can I calibrate my approach so 
that I can still make you 

724
00:35:00,150 --> 00:35:02,102
effective or maybe two different
means. 

725
00:35:03,285 --> 00:35:05,580
Yeah, it's a growth part for me,
I think, yeah. 

726
00:35:06,295 --> 00:35:08,635
In my mind as well, I mean, 
like, it's always hard, right? 

727
00:35:08,665 --> 00:35:11,415
Whether you wanna go hard on 
certain things or whether you 

728
00:35:11,415 --> 00:35:15,134
kind of like, you know, let the 
team kind of like solve it, set 

729
00:35:15,134 --> 00:35:18,312
the standard by themselves. 
It's always very tough because 

730
00:35:18,312 --> 00:35:22,085
people these days, you know, 
talk about, I don't know, like 

731
00:35:22,085 --> 00:35:25,813
giving them trust, psychological
safety, you know, they are not 

732
00:35:25,813 --> 00:35:28,663
like a physical sports team. 
They're more like a knowledge 

733
00:35:28,663 --> 00:35:30,802
worker. 
It's always a tough balance 

734
00:35:30,802 --> 00:35:34,471
approach, I would say. 
But when you say you had some 

735
00:35:34,471 --> 00:35:37,710
failure stories as well, right? 
Like what kind of things that 

736
00:35:37,710 --> 00:35:41,068
maybe don't work well with this 
tough approach towards software 

737
00:35:41,068 --> 00:35:43,849
engineering team, right? 
Like what, what exactly that you

738
00:35:43,849 --> 00:35:46,581
learned from those experience 
that you kind of like tweaked 

739
00:35:46,581 --> 00:35:50,009
along the way? 
Yeah, so I think sometimes when 

740
00:35:50,009 --> 00:35:53,967
you take like a tough approach, 
the signal to noise ratio gets 

741
00:35:53,967 --> 00:35:57,476
warped, because people anchor a 
lot on how you say things as 

742
00:35:57,476 --> 00:36:01,456
opposed to what you say. 
So the tough approach works 

743
00:36:01,456 --> 00:36:03,813
really well when... 
You know it's like radical 

744
00:36:03,813 --> 00:36:06,801
candor to some degree. 
I know the tough approach and 

745
00:36:06,801 --> 00:36:08,972
radical candor doesn't 
necessarily mean the same thing,

746
00:36:08,972 --> 00:36:11,210
but it overlaps. 
Because if you're being very 

747
00:36:11,210 --> 00:36:13,246
frank with someone, it means 
that if you're like frustrated 

748
00:36:13,246 --> 00:36:15,933
with them, you kinda share the 
fact that the thing is 

749
00:36:15,933 --> 00:36:17,729
frustrating you. 
And that's sometimes is seen as 

750
00:36:17,729 --> 00:36:21,173
tough. 
Now, I think for it to be really

751
00:36:21,173 --> 00:36:24,323
effective, you firstly need to 
have established enough trust 

752
00:36:24,323 --> 00:36:28,406
that the person who's receiving 
the feedback knows that your 

753
00:36:28,406 --> 00:36:31,549
intention is good. 
Your intention is to help them 

754
00:36:31,549 --> 00:36:35,291
on their project. 
If you don't have that 

755
00:36:35,291 --> 00:36:39,575
foundation in place, anything 
you say in this like tough 

756
00:36:39,575 --> 00:36:42,656
approach will be taken as an 
attack to them. 

757
00:36:42,986 --> 00:36:45,136
So you don't have that 
counterbalance. 

758
00:36:45,826 --> 00:36:49,586
Now to build that trust though, 
it takes investment and time. 

759
00:36:49,646 --> 00:36:54,206
So, it's a high, it requires 
high bandwidth communication. 

760
00:36:54,750 --> 00:36:58,686
So I think it's more effective 
within like a smaller circle of 

761
00:36:58,686 --> 00:37:01,788
people who you can establish 
that personal connection with 

762
00:37:01,788 --> 00:37:05,418
and that trust with, as opposed 
to like a more broader, wider 

763
00:37:05,418 --> 00:37:07,284
group. 
Because for the wider group, 

764
00:37:07,284 --> 00:37:10,749
they're only gonna hear the 
times when you make a lot of 

765
00:37:10,749 --> 00:37:13,465
noise. 
They're not gonna be there when 

766
00:37:13,465 --> 00:37:15,978
you're like in the one-on-one, 
kind of like explain, okay, this

767
00:37:15,978 --> 00:37:18,660
is why this didn't work out, or 
this is why I'm frustrated this.

768
00:37:18,660 --> 00:37:21,870
Or like you tell me like why you
think that way, right? 

769
00:37:21,870 --> 00:37:24,896
Like it's just not there. 
So that's where I think the 

770
00:37:24,896 --> 00:37:27,881
calibration is required, like 
just being more conscious of 

771
00:37:27,881 --> 00:37:30,271
what message are we getting 
across? 

772
00:37:30,271 --> 00:37:33,887
Is it the most effective way? 
Also, I mean, I'll be frank with

773
00:37:33,887 --> 00:37:36,690
you, like you know, it's, I 
think it's a lot about ego as 

774
00:37:36,690 --> 00:37:38,860
well. 
Sometimes being tough and like, 

775
00:37:38,860 --> 00:37:42,177
you know, grandstanding is, is 
an ego thing as well. 

776
00:37:42,267 --> 00:37:45,777
And I'm trying to make sure that
I don't fall into that trap 

777
00:37:45,777 --> 00:37:49,407
where I myself wanna say this 
this way because it makes me 

778
00:37:49,407 --> 00:37:53,321
feel good. 
Trying more to focus more like, 

779
00:37:53,321 --> 00:37:57,136
okay, what is the most effective
thing to be able to get the 

780
00:37:57,136 --> 00:37:59,345
outcome we need? 
And that takes effort and 

781
00:37:59,345 --> 00:38:03,071
discipline too. 
So yeah, definitely, I think for

782
00:38:03,071 --> 00:38:06,499
those leaders who are also like 
kind of like contemplating the 

783
00:38:06,499 --> 00:38:09,121
approach themselves, because I 
think you learn it throughout 

784
00:38:09,121 --> 00:38:11,629
your experience, right? 
There's no doubt like what you 

785
00:38:11,629 --> 00:38:12,799
mentioned, you learn from 
childhood. 

786
00:38:12,799 --> 00:38:15,385
Maybe after certain years of 
your career, you also learn 

787
00:38:15,385 --> 00:38:17,479
along the way. 
Maybe you meet somebody who 

788
00:38:17,479 --> 00:38:19,279
teach you different way of doing
things. 

789
00:38:19,639 --> 00:38:21,499
Obviously there's no right or 
wrong answer. 

790
00:38:21,769 --> 00:38:24,699
It's highly contextual as well. 
Maybe sometimes the team needs 

791
00:38:24,699 --> 00:38:26,699
this kind of more disciplined 
type approach. 

792
00:38:27,219 --> 00:38:30,503
And I like that you mentioned, 
you know, leaders have the 

793
00:38:30,503 --> 00:38:33,097
responsibility to have this high
bandwidth communication and 

794
00:38:33,097 --> 00:38:36,735
building trust, right, because 
you can't just, you know, using 

795
00:38:36,735 --> 00:38:39,039
tough approach to get certain 
results. 

796
00:38:39,424 --> 00:38:41,924
I think building the trust is 
definitely still one thing. 

797
00:38:42,604 --> 00:38:45,734
And the ego aspects, I think in 
every kind of debate within 

798
00:38:45,734 --> 00:38:47,264
software engineering team, 
right? 

799
00:38:47,624 --> 00:38:51,446
I think there's always some kind
of ego, I feel, like maybe this 

800
00:38:51,446 --> 00:38:54,216
tech stack that you really love 
or maybe this approach you 

801
00:38:54,216 --> 00:38:56,580
really love or whatever 
solutions that you think is the 

802
00:38:56,580 --> 00:38:59,213
best way. 
I think, yeah, putting ego aside

803
00:38:59,213 --> 00:39:02,045
is definitely very important. 
I mean, it works not just within

804
00:39:02,045 --> 00:39:04,599
software engineering team, I 
think in the real life as well, 

805
00:39:04,599 --> 00:39:06,500
I would say. 
A hundred percent. 

806
00:39:07,230 --> 00:39:10,326
One aspect that I see that I 
think in your career that I 

807
00:39:10,326 --> 00:39:13,033
think very unique and maybe 
interesting to learn from is 

808
00:39:13,033 --> 00:39:16,952
like, there was a point in time 
where you go back working to 

809
00:39:16,952 --> 00:39:19,265
Google, right? 
Um, so I've seen your career, 

810
00:39:19,265 --> 00:39:22,352
you have been like, I don't 
know, CTO, you know, senior vice

811
00:39:22,352 --> 00:39:25,780
president, you know, managing 
big, large engineering team. 

812
00:39:25,780 --> 00:39:29,578
And then you took a decision to 
actually go back as an IC, 

813
00:39:29,578 --> 00:39:30,991
right? 
When you go back to Google. 

814
00:39:31,111 --> 00:39:32,506
And that was kinda like a short 
stint. 

815
00:39:33,151 --> 00:39:35,824
So, and for many people, 
especially leaders who have 

816
00:39:35,824 --> 00:39:39,808
been, you know, in the 15 or 10 
year of their career as a 

817
00:39:39,808 --> 00:39:42,689
manager, right? 
Sometimes they wish to go back 

818
00:39:42,689 --> 00:39:46,017
as an IC. 
But sometimes they have so many 

819
00:39:46,017 --> 00:39:49,015
different thoughts. 
Maybe if you can outline to us, 

820
00:39:49,015 --> 00:39:51,671
like what was your thought 
process back then? 

821
00:39:51,881 --> 00:39:54,741
Was it difficult for you to move
back from management to IC? 

822
00:39:55,191 --> 00:39:56,921
And yeah, maybe we start from 
there. 

823
00:39:57,141 --> 00:40:00,290
Sure, no, absolutely! 
It's, it's a question that I've 

824
00:40:00,290 --> 00:40:03,358
gotten before. 
Okay, so firstly, for me, was 

825
00:40:03,358 --> 00:40:06,300
first and foremost an 
opportunity to work at Google. 

826
00:40:07,104 --> 00:40:09,200
I know you, Henry, you've worked
at Google as well. 

827
00:40:09,948 --> 00:40:14,326
I wanted to experience, take a 
view of inside the Googleplex, 

828
00:40:14,326 --> 00:40:17,788
okay, what it is. 
Unfortunately, I joined during 

829
00:40:17,788 --> 00:40:20,968
COVID, so it wasn't the best 
experience. 

830
00:40:20,988 --> 00:40:24,572
But all kudos to Google. 
I think they, they did a really 

831
00:40:24,572 --> 00:40:26,388
good job with onboarding. 
They went out their way. 

832
00:40:27,065 --> 00:40:30,255
And that was great. 
So the opportunity to be an IC 

833
00:40:30,255 --> 00:40:33,569
and kind of like bootstrap and 
learn like, okay, what would it 

834
00:40:33,569 --> 00:40:37,059
mean if I was like a software 
engineer or like a, back then 

835
00:40:37,059 --> 00:40:40,709
the role of solution architect 
within Google, that was really 

836
00:40:40,709 --> 00:40:43,824
alluring because I wanted to 
learn again. 

837
00:40:43,824 --> 00:40:46,804
I wanted to see how these big 
companies do it. 

838
00:40:46,804 --> 00:40:49,804
They've done fantastic stuff. 
Also, just the pedigree, right? 

839
00:40:49,804 --> 00:40:52,994
Like the ability to kind of like
look through code commits from 

840
00:40:52,994 --> 00:40:56,905
Jeff Dean or Sanjay Ghemawat or 
like, you know, look through the

841
00:40:56,905 --> 00:40:59,911
code for Bigtable. 
I didn't understand a lot of it,

842
00:40:59,911 --> 00:41:02,095
but it was just the experience 
of being immersed in that. 

843
00:41:02,095 --> 00:41:07,250
So the IC role for me was almost
secondary to the opportunity. 

844
00:41:07,940 --> 00:41:12,763
Having said that, I think, the 
IC role was also an opportunity 

845
00:41:12,763 --> 00:41:14,994
for me to kind of test my own 
skills. 

846
00:41:14,994 --> 00:41:19,531
Like, while I had been leading 
teams all the way up to like 

847
00:41:19,531 --> 00:41:23,055
2014, 2015, 2016, I was still 
kind of working on code as well.

848
00:41:23,085 --> 00:41:24,705
Cause the team wasn't that 
large. 

849
00:41:24,705 --> 00:41:30,727
We were like, when I was in 
Indonesia, and we were probably 

850
00:41:30,727 --> 00:41:35,740
2014-15, it was like a 70 person
team, 70-80 person team. 

851
00:41:35,800 --> 00:41:37,600
So still below the Dunbar 
number. 

852
00:41:38,380 --> 00:41:40,900
I could go in the morning, we'll
do stand up. 

853
00:41:41,635 --> 00:41:43,285
I'll spend a couple hours pair 
programming. 

854
00:41:43,945 --> 00:41:46,885
So we, with someone on the team.
It was very common. 

855
00:41:46,885 --> 00:41:50,905
Tommy did the same thing. 
Like we spent more time, we 

856
00:41:50,905 --> 00:41:54,553
spent very little time in 
meetings because the team was 

857
00:41:54,553 --> 00:41:56,092
firstly self-organized. 
We had the different touch 

858
00:41:56,092 --> 00:41:58,183
points a day. 
We'll have a standup in the 

859
00:41:58,183 --> 00:41:59,490
morning, teams, company wide 
stand up. 

860
00:41:59,490 --> 00:42:00,945
So we'll have all the engineers 
kind of do it. 

861
00:42:01,215 --> 00:42:03,465
Then we'll have a team stand up 
and then we'll break. 

862
00:42:03,495 --> 00:42:07,455
And uh, you know, I was lucky 
to, Emtek was relatively hands 

863
00:42:07,455 --> 00:42:10,831
off in a good way. 
They kind of just let us do our 

864
00:42:10,831 --> 00:42:12,527
stuff. 
We had product team members who 

865
00:42:12,527 --> 00:42:15,575
would work more closely with the
business and we were somewhat 

866
00:42:15,575 --> 00:42:17,941
firewalled from that for, which 
was great. 

867
00:42:18,181 --> 00:42:20,900
We didn't have too many meeting.
So I was very involved in the 

868
00:42:20,900 --> 00:42:24,241
code up to like maybe 2014, 
2015, I worked very close. 

869
00:42:24,301 --> 00:42:27,126
Before we hired more senior 
people in the DevOps team, I was

870
00:42:27,126 --> 00:42:29,761
practically like the tech lead 
for the DevOps team. 

871
00:42:30,694 --> 00:42:33,841
And then only like 2017, 2018, 
where I really kind of rolled 

872
00:42:33,841 --> 00:42:37,729
off code again and was more, not
actively writing code, but 

873
00:42:37,729 --> 00:42:40,357
definitely involved in some of 
the technical aspects, which is 

874
00:42:40,357 --> 00:42:41,977
kind of the role I play today as
well. 

875
00:42:42,157 --> 00:42:45,742
So going back to an IC role was 
an opportunity to kind of 

876
00:42:45,742 --> 00:42:48,814
reconnect with some of that, 
especially given it was solution

877
00:42:48,814 --> 00:42:51,337
architecture within Google 
Cloud, it was like DevOps and 

878
00:42:51,337 --> 00:42:53,928
stuff like that, which is an 
area that is close to my heart 

879
00:42:53,928 --> 00:42:55,257
and something I thought would be
good at. 

880
00:42:55,883 --> 00:42:58,131
So that was the hypothesis going
in. 

881
00:42:59,121 --> 00:43:01,817
I think couple of things 
happened, some internal to 

882
00:43:01,817 --> 00:43:05,097
Google which is not important, 
but I think more externally, 

883
00:43:05,097 --> 00:43:08,791
what I quickly realized is that 
I wasn't getting the dopamine 

884
00:43:08,791 --> 00:43:12,003
hit of being able to drive a lot
more outcome. 

885
00:43:13,443 --> 00:43:19,077
I realized that my ability to 
kind of scale my impact as an IC

886
00:43:19,077 --> 00:43:23,175
was very limited. 
You know, I would literally go 

887
00:43:23,175 --> 00:43:25,787
around looking for projects to 
kind of contribute to, and there

888
00:43:25,787 --> 00:43:29,343
were some projects that I could,
and it was great and it was fun.

889
00:43:29,883 --> 00:43:33,374
But I just very quickly realized
that Google's a large 

890
00:43:33,374 --> 00:43:36,235
organization and they have swim 
lanes for people and they're 

891
00:43:36,235 --> 00:43:38,921
very happy for people. 
Or at least not, I wouldn't say 

892
00:43:38,921 --> 00:43:41,163
very happy, but the-sense was 
people wanted to stay in their 

893
00:43:41,163 --> 00:43:44,514
swim lanes. 
So in my swim lane, I had only 

894
00:43:44,514 --> 00:43:46,863
so much to do, so much impact to
drive. 

895
00:43:46,923 --> 00:43:50,553
And my manager was okay, my 
managers were happy, but I 

896
00:43:50,553 --> 00:43:54,333
wasn't enjoying my work as much.
So when Suthen got in touch with

897
00:43:54,333 --> 00:43:56,393
me again, he said like, okay, 
there's this interesting 

898
00:43:56,393 --> 00:43:57,943
opportunity. 
We really want to do this. 

899
00:43:57,943 --> 00:44:00,202
I think you'd be a good person. 
I had interviewed at Grab 

900
00:44:00,202 --> 00:44:02,841
before. 
He said, you know, if you're 

901
00:44:02,841 --> 00:44:05,223
keen, you have to come now. 
He said, like, you know, this 

902
00:44:05,223 --> 00:44:07,383
role is gonna be open only for a
certain period of time. 

903
00:44:08,583 --> 00:44:12,243
I was like, damnit, I was, we 
were just to have my second kid.

904
00:44:12,243 --> 00:44:14,403
So the Google paternity would've
been great. 

905
00:44:14,953 --> 00:44:19,015
Grab's paternity back then was 
only two weeks, but he was like,

906
00:44:19,015 --> 00:44:20,647
no. 
Uh, I mean, he was nice about 

907
00:44:20,647 --> 00:44:22,651
it. 
He wanted to see how to make it 

908
00:44:22,651 --> 00:44:25,835
works and I was like, I was 
definitely keen on working on 

909
00:44:25,835 --> 00:44:28,109
bigger problems. 
And for me, bigger problems. 

910
00:44:28,919 --> 00:44:31,469
Again, it's more a reflection of
my skillset. 

911
00:44:31,499 --> 00:44:33,629
I'm not saying an IC can't drive
bigger impact. 

912
00:44:34,199 --> 00:44:37,349
It's just, you know, my ability,
if I look at my strengths. 

913
00:44:37,349 --> 00:44:38,849
I think Peter Drucker says it, 
right? 

914
00:44:38,849 --> 00:44:42,201
Like, sometimes people focus a 
lot more on their growth areas 

915
00:44:42,201 --> 00:44:43,834
or weaknesses. 
But actually in your career, 

916
00:44:43,834 --> 00:44:45,949
especially in your latter part, 
you should focus on your 

917
00:44:45,949 --> 00:44:48,229
strengths and drive that up 
more. 

918
00:44:48,529 --> 00:44:50,721
So kind of some of that 
thinking, like I realized like 

919
00:44:50,721 --> 00:44:54,167
I'm probably stronger as like a 
team lead, tech lead type, 

920
00:44:54,167 --> 00:44:57,024
engineering leader type role. 
And there was a big opportunity 

921
00:44:57,024 --> 00:44:58,729
at Grab at that point. 
So I went. 

922
00:44:58,819 --> 00:45:00,933
Took the plunge. 
So definitely, it's very 

923
00:45:00,933 --> 00:45:02,899
interesting to learn from your 
experience, right? 

924
00:45:02,899 --> 00:45:05,763
Because there are some people 
who just really love hands-on 

925
00:45:05,763 --> 00:45:08,455
and maybe can drive impact, you 
know, building a scalable 

926
00:45:08,455 --> 00:45:09,997
solution or whatever that is, 
right? 

927
00:45:10,027 --> 00:45:12,997
Unique, innovative, and that 
could still work, right? 

928
00:45:12,997 --> 00:45:16,747
But for some people who are born
as leaders, I would say, right? 

929
00:45:17,077 --> 00:45:20,395
So I think the dopamine hit is 
to actually drive impact through

930
00:45:20,395 --> 00:45:22,621
people, right? 
So I think that is maybe one 

931
00:45:22,621 --> 00:45:25,461
thing that we can learn from 
this story that you just shared,

932
00:45:25,461 --> 00:45:27,274
right? 
See what makes you tick. 

933
00:45:27,484 --> 00:45:31,971
I mean, if you love making an 
impact through people you know, 

934
00:45:31,971 --> 00:45:34,704
creating, I don't know, business
objectives or outcomes through, 

935
00:45:34,704 --> 00:45:37,648
you know, leading teams. 
I think there could be aspects 

936
00:45:37,648 --> 00:45:40,000
that maybe suits you well in the
management. 

937
00:45:40,634 --> 00:45:45,361
And I think one thing that you 
mentioned about leaders, and 

938
00:45:45,511 --> 00:45:47,666
wanting to be hands-on again, 
right? 

939
00:45:48,206 --> 00:45:51,742
And I know these days, it is the
fear of many leaders, including 

940
00:45:51,742 --> 00:45:54,691
me, right, with the, again, 
coming back to AI, right? 

941
00:45:55,371 --> 00:45:59,004
So a lot of more stuff now can 
be leveraged through AI. 

942
00:45:59,364 --> 00:46:03,714
And, and that also means like 
teams getting smaller and maybe 

943
00:46:03,714 --> 00:46:08,076
you are required even to use AI 
to actually do more hands-on 

944
00:46:08,076 --> 00:46:10,770
work. 
And this goes back to some of 

945
00:46:10,770 --> 00:46:13,325
your rationale trying to be 
hands-on, right, in your career.

946
00:46:13,355 --> 00:46:16,347
So do you think it's really 
important for leaders these days

947
00:46:16,347 --> 00:46:20,693
to come back to being hands on 
and, you know, learn all these, 

948
00:46:20,693 --> 00:46:23,410
you know, recent technologies, 
advancements, and those stuff? 

949
00:46:24,335 --> 00:46:27,891
It's a really good question. 
It's a question that I think I 

950
00:46:27,891 --> 00:46:30,532
agree with you. 
It's becoming even more relevant

951
00:46:30,532 --> 00:46:34,946
in this AI, you know, AI infused
software engineering career now.

952
00:46:35,619 --> 00:46:40,839
My philosophy's always been that
you wouldn't get trained by a 

953
00:46:41,049 --> 00:46:44,499
coach, a tennis coach that 
doesn't himself play or herself 

954
00:46:44,499 --> 00:46:48,578
play tennis. 
I think that, as a basic rule of

955
00:46:48,578 --> 00:46:50,930
thumb, applies to technical 
management as well. 

956
00:46:51,860 --> 00:46:55,130
Now, the tennis coach might not 
be as good as you in tennis, but

957
00:46:55,130 --> 00:46:57,740
they know techniques. 
They've been around long enough.

958
00:46:57,770 --> 00:47:00,650
They've seen enough plays, and 
they're also kind of like 

959
00:47:00,650 --> 00:47:03,860
continuing to watch games to see
new approaches and techniques. 

960
00:47:03,860 --> 00:47:05,300
They're trying it out on their 
own. 

961
00:47:05,528 --> 00:47:08,178
They're probably not gonna go 
for the next grand slam 

962
00:47:08,178 --> 00:47:10,361
tournament, but they're there. 
They're plugged into it. 

963
00:47:10,961 --> 00:47:13,301
I think that same thing applies 
to technical management. 

964
00:47:14,026 --> 00:47:17,584
You have to be plugged in. 
You cannot just be, maybe it's 

965
00:47:17,584 --> 00:47:21,128
more the sphere or the space 
that I've been focused on, which

966
00:47:21,128 --> 00:47:23,603
definitely has been more on the 
platform, infrastructure side. 

967
00:47:24,023 --> 00:47:27,167
But I just don't believe, I 
think it's hard for someone just

968
00:47:27,167 --> 00:47:30,196
to be a 'manager' manager. 
Like there's a role to play, 

969
00:47:30,196 --> 00:47:33,082
don't get me wrong. 
Especially when it's a lot of 

970
00:47:33,082 --> 00:47:35,910
like business contacts and like 
customer contacts, but then 

971
00:47:35,910 --> 00:47:39,075
you're kind of playing a role 
almost like a product manager, 

972
00:47:39,075 --> 00:47:42,379
engineering manager. 
And maybe for some type of 

973
00:47:42,379 --> 00:47:44,080
problem, some type of companies 
that works. 

974
00:47:44,673 --> 00:47:48,601
But generally, my preference is 
always to find someone who's 

975
00:47:48,601 --> 00:47:52,212
more of a technical manager and 
help them build those other 

976
00:47:52,212 --> 00:47:54,577
skills. 
Especially if there's already 

977
00:47:54,577 --> 00:47:57,513
like a job function or role for 
someone to be product manager. 

978
00:47:57,573 --> 00:47:59,583
Cause otherwise, it just ends up
like overlapping. 

979
00:47:59,733 --> 00:48:01,713
Or they become like project 
managers, right? 

980
00:48:01,713 --> 00:48:04,586
They just kind of, I wouldn't 
even say, I think TPMs do more 

981
00:48:04,586 --> 00:48:06,216
interesting work, but like you 
know, like, yeah. 

982
00:48:06,216 --> 00:48:09,215
Just like a project manager and 
I feel that's also not 

983
00:48:09,215 --> 00:48:11,365
productive. 
So then the question is like, 

984
00:48:11,365 --> 00:48:13,147
okay, how do you find the time, 
right? 

985
00:48:13,147 --> 00:48:14,317
Especially if you're managing 
people. 

986
00:48:14,377 --> 00:48:15,769
That's a good question. 
And I agree. 

987
00:48:15,769 --> 00:48:18,229
It's a real challenge if you 
want to do it justice. 

988
00:48:18,339 --> 00:48:21,716
It's a real challenge. 
You know, at Grab, we have this,

989
00:48:21,716 --> 00:48:24,661
we have like in our career 
ladder, we kind of say that, 

990
00:48:24,661 --> 00:48:27,135
okay, managers, like line 
managers should be able to 

991
00:48:27,135 --> 00:48:29,145
operate the ICs on their team as
well. 

992
00:48:29,685 --> 00:48:33,223
Meaning that doesn't mean they 
contribute code actively on a 

993
00:48:33,223 --> 00:48:36,847
daily basis, but they can 
contribute at that level of 

994
00:48:36,847 --> 00:48:38,781
capacity. 
It's a bit of a controversial 

995
00:48:38,781 --> 00:48:41,251
statement, we've gotten all sort
feedback on it, but I think it's

996
00:48:41,251 --> 00:48:43,835
still the right framing. 
Now what does that practically 

997
00:48:43,835 --> 00:48:45,845
mean for a manager at that 
level? 

998
00:48:46,325 --> 00:48:50,175
I think it means like, you know,
ensuring and working with your 

999
00:48:50,175 --> 00:48:53,231
manager to ensure that you have 
enough time set aside to be able

1000
00:48:53,231 --> 00:48:56,195
to review code, to participate 
in RFCs and all that. 

1001
00:48:56,195 --> 00:48:58,415
And if you find yourself just 
doing people management 

1002
00:48:58,415 --> 00:49:00,095
throughout the day, that's 
broken. 

1003
00:49:00,665 --> 00:49:04,327
Like I feel I see some people 
who like have this accelerated 

1004
00:49:04,327 --> 00:49:07,937
part into management. 
I mean, and typically these 

1005
00:49:07,937 --> 00:49:10,415
people are, these folks are very
good. 

1006
00:49:11,330 --> 00:49:14,552
But they spend, they don't spend
enough time doing enough 

1007
00:49:14,552 --> 00:49:16,822
hands-on work. 
So if you think about your 

1008
00:49:16,822 --> 00:49:19,513
career, like, okay, if you start
working like, I don't know, like

1009
00:49:19,513 --> 00:49:22,631
22, 23, right? 
And let's say, let's say, okay, 

1010
00:49:22,631 --> 00:49:25,611
you're lucky you can cash out at
the age of 50, right? 

1011
00:49:25,821 --> 00:49:27,441
You're talking about 27, 30 
years. 

1012
00:49:28,071 --> 00:49:32,364
If you've only spent like, let, 
let's say 30, maybe 35 years, 

1013
00:49:32,364 --> 00:49:35,256
right? 
If you are a junior engineer and

1014
00:49:35,256 --> 00:49:37,739
then you are becoming an 
engineering manager within like 

1015
00:49:37,739 --> 00:49:41,620
the first five to seven years of
a career, how much time have you

1016
00:49:41,620 --> 00:49:44,804
spent really building any type 
of foundational knowledge? 

1017
00:49:45,404 --> 00:49:47,645
I just feel it's a very... 
And then you're gonna spend all 

1018
00:49:47,645 --> 00:49:49,241
the other time just as a 
manager. 

1019
00:49:49,991 --> 00:49:52,054
I just don't think it's the 
right way to approach these 

1020
00:49:52,054 --> 00:49:53,281
things. 
I think people should dedicate 

1021
00:49:53,281 --> 00:49:56,981
more time kind of building out 
the IC skills or in situations 

1022
00:49:56,981 --> 00:49:59,423
where. 
So like I, my career was a bit 

1023
00:49:59,423 --> 00:50:01,711
like that. 
Like, you know, I kind of spent 

1024
00:50:01,711 --> 00:50:04,607
six, seven years as an IC, but 
then kind of very quickly 

1025
00:50:04,607 --> 00:50:06,942
shifted into like leading a 
small team and stuff like that. 

1026
00:50:06,971 --> 00:50:09,706
I think that's okay. 
If the team small enough that 

1027
00:50:09,706 --> 00:50:11,351
you can still be hands-on, 
you're contributing. 

1028
00:50:11,351 --> 00:50:13,978
So you're doing a bit of both. 
It's a bit like a tech lead 

1029
00:50:13,978 --> 00:50:16,871
manager type of setup, which 
lots of people, and at Grab, we 

1030
00:50:16,871 --> 00:50:19,905
don't really promote it as well.
And there are good reasons why 

1031
00:50:19,905 --> 00:50:22,792
it's not a great model, but if 
you find yourself in that setup,

1032
00:50:22,792 --> 00:50:26,181
and I think it's still okay. 
So how do you kind of create 

1033
00:50:26,181 --> 00:50:28,945
this opportunities, where you 
have enough bandwidth to still 

1034
00:50:28,945 --> 00:50:32,025
spend time being close? 
And then also I think picking 

1035
00:50:32,025 --> 00:50:35,903
problems which allow you to 
continue to grow in that 

1036
00:50:35,903 --> 00:50:38,776
direction. 
So you could be a manager, but 

1037
00:50:38,776 --> 00:50:42,233
you pick technical domains or 
scopes that allow you to get 

1038
00:50:42,233 --> 00:50:45,986
more technical. 
Like pushes you and you apply 

1039
00:50:45,986 --> 00:50:49,894
your technical aspect of your 
skillset in that domain. 

1040
00:50:49,894 --> 00:50:51,834
And you don't just focus on the 
people management side. 

1041
00:50:52,684 --> 00:50:56,560
I feel that's really important. 
What about those roles that is 

1042
00:50:56,560 --> 00:51:01,429
really high up, like CTO, SVP, 
do you still think they need to 

1043
00:51:01,429 --> 00:51:04,170
be hands-on picking up for 
certain problems to chase, 

1044
00:51:04,170 --> 00:51:06,094
technically, not just like 
people management? 

1045
00:51:06,379 --> 00:51:09,532
I again, I think, you know, I 
don't wanna speak on behalf of 

1046
00:51:09,532 --> 00:51:12,320
everyone. 
I definitely I think different 

1047
00:51:12,320 --> 00:51:16,074
roles, uh, different companies 
vary in terms of type of 

1048
00:51:16,074 --> 00:51:18,514
problems they want their CTOs or
their VPs to solve. 

1049
00:51:18,904 --> 00:51:23,085
Having said that, I think it 
depends on the individual again,

1050
00:51:23,085 --> 00:51:26,356
right, and the problem. 
So a lot of it will come down to

1051
00:51:26,356 --> 00:51:28,351
the company. 
So for example, when I was hired

1052
00:51:28,351 --> 00:51:30,127
into BUKA, there was a very 
clear mandate. 

1053
00:51:30,157 --> 00:51:33,347
Okay, Mohan, you come in, we 
need to migrate from on-prem to 

1054
00:51:33,347 --> 00:51:37,291
the cloud. 
So my mandate coming in was the 

1055
00:51:37,291 --> 00:51:39,532
technical mandate. 
Okay, we need to get this stuff 

1056
00:51:39,532 --> 00:51:41,731
done. 
And you know, there was this new

1057
00:51:41,731 --> 00:51:45,989
cloud that nobody had ever used.
It was a team that was used to 

1058
00:51:45,989 --> 00:51:47,591
only managing stuff on-prem 
before. 

1059
00:51:48,231 --> 00:51:51,342
Okay, so there was a, there was 
both a social aspect, but there 

1060
00:51:51,342 --> 00:51:53,787
was also a technical aspect that
needed to be brought together 

1061
00:51:53,787 --> 00:51:56,231
and solved. 
So I think CTOs or VPs, that's 

1062
00:51:56,231 --> 00:51:58,151
kind of the type of problems 
you're solving. 

1063
00:51:58,151 --> 00:52:00,588
So you need to have still, you 
need to have enough technical 

1064
00:52:00,588 --> 00:52:02,022
knowledge. 
You need to be able to answer 

1065
00:52:02,022 --> 00:52:03,941
the question, like, okay, what 
type of skills that maybe before

1066
00:52:03,941 --> 00:52:06,581
this my team had, and now 
they're gonna need to have? 

1067
00:52:06,791 --> 00:52:09,311
What does that mean practically?
Because otherwise people can 

1068
00:52:09,311 --> 00:52:10,811
come and just like gas you, 
right? 

1069
00:52:10,811 --> 00:52:12,401
They'll come and say like, oh, 
this is too hard. 

1070
00:52:12,401 --> 00:52:14,955
It's gonna take a, we need to 
hire five more people to be able

1071
00:52:14,955 --> 00:52:17,328
to do this. 
And you're gonna go like, how do

1072
00:52:17,328 --> 00:52:18,894
I challenge question this 
person, right? 

1073
00:52:19,194 --> 00:52:21,199
Yeah. 
So I think within like that, 

1074
00:52:21,199 --> 00:52:25,016
those capacities, you need to be
close enough to be able to put 

1075
00:52:25,016 --> 00:52:27,721
forward a credible 
counter-argument or review what 

1076
00:52:27,721 --> 00:52:31,933
is being shared with you and not
just kind of rely on proxying 

1077
00:52:31,933 --> 00:52:34,020
out. 
Cause you are a, I mean, there's

1078
00:52:34,020 --> 00:52:36,714
a T in the CTO, right? 
It's technical, technology. 

1079
00:52:36,714 --> 00:52:39,054
Obviously, that's got, that's 
gonna mean something. 

1080
00:52:40,147 --> 00:52:42,492
For some people, I don't know 
like you mentioned a few 

1081
00:52:42,492 --> 00:52:45,115
practice that I don't know 
whether you have, you are still 

1082
00:52:45,115 --> 00:52:47,889
doing it now, like reviewing 
code, reviewing PR, for example,

1083
00:52:47,889 --> 00:52:51,235
uh, pair programming with them, 
or maybe chasing a certain 

1084
00:52:51,235 --> 00:52:53,125
particular bug during incidents,
right? 

1085
00:52:53,245 --> 00:52:55,885
Some people will treat this, 
especially if you are like a 

1086
00:52:55,885 --> 00:52:58,915
higher up management, treat it 
as a micromanagement. 

1087
00:52:59,460 --> 00:53:01,375
Yeah. 
So what is your view on this? 

1088
00:53:01,555 --> 00:53:05,248
No, I, listen, I think if you 
have a structure in place. 

1089
00:53:05,248 --> 00:53:08,598
Let's say, for example, you're a
CTO of a 400 person org. 

1090
00:53:09,108 --> 00:53:12,708
Yeah, I don't expect you to be 
sitting down reviewing the MR. 

1091
00:53:12,708 --> 00:53:18,292
But I expect you to be sitting 
in the team post-mortem review 

1092
00:53:18,292 --> 00:53:23,777
and providing feedback. 
I expect you to make sure that 

1093
00:53:23,777 --> 00:53:26,923
your delegates are ensuring that
the bugs are being closed. 

1094
00:53:27,572 --> 00:53:31,964
And I expect you to understand 
the root causes, maybe through 

1095
00:53:31,964 --> 00:53:36,128
the postmodern review of why 
maybe, oh, this thing has been 

1096
00:53:36,128 --> 00:53:39,032
happening again and again. 
Is this because this team has 

1097
00:53:39,032 --> 00:53:40,442
taken on too much technical 
debt? 

1098
00:53:40,682 --> 00:53:43,052
Like forming more strategic 
views on top of this. 

1099
00:53:43,052 --> 00:53:48,362
But like this strategic lens can
only be built on the details. 

1100
00:53:49,174 --> 00:53:51,924
So you, you, you, you're 
absolutely right. 

1101
00:53:51,944 --> 00:53:53,906
You shouldn't be micromanaging, 
but that doesn't mean you don't 

1102
00:53:53,906 --> 00:53:56,626
know the details. 
You know, AWS has this 

1103
00:53:56,626 --> 00:54:00,342
leadership principles. 
My favorite of all the 

1104
00:54:00,342 --> 00:54:02,796
leadership principles is leaders
dive deep. 

1105
00:54:03,520 --> 00:54:07,690
I feel that that really sets the
tone the type of leaders, right?

1106
00:54:07,690 --> 00:54:11,362
So the leaders dive deep, 
they're auditing, they are 

1107
00:54:11,362 --> 00:54:13,630
balancing like data with 
anecdotes. 

1108
00:54:13,840 --> 00:54:17,701
They're doing the investigation.
I agree with you, you shouldn't 

1109
00:54:17,701 --> 00:54:19,564
be micromanaging. 
But there's lots that you can be

1110
00:54:19,564 --> 00:54:21,493
doing that's not within the 
realm of micromanagement. 

1111
00:54:21,763 --> 00:54:25,825
Especially if you have people 
who are already empowered to do,

1112
00:54:25,825 --> 00:54:28,593
you know, some of the more 
ground level details. 

1113
00:54:29,316 --> 00:54:32,676
And by doing it this way, you're
setting the tone for a team. 

1114
00:54:32,676 --> 00:54:34,812
So you're saying like, this is 
important enough for the CTO to 

1115
00:54:34,812 --> 00:54:37,681
be aware of. 
And I think that really matters 

1116
00:54:37,681 --> 00:54:39,403
to setting the right engineering
culture. 

1117
00:54:39,403 --> 00:54:42,207
Yeah, and it goes back to what 
you mentioned, setting the right

1118
00:54:42,207 --> 00:54:44,423
bar, right? 
Because if you actually don't 

1119
00:54:44,423 --> 00:54:46,673
understand the technical 
details, definitely, the bar 

1120
00:54:46,673 --> 00:54:49,168
will be just low, right? 
And I like what you mentioned, 

1121
00:54:49,168 --> 00:54:51,668
right? 
Comparing data vs anecdotes. 

1122
00:54:51,928 --> 00:54:54,810
Sometimes when you are higher 
up, you, you can just hear 

1123
00:54:54,810 --> 00:54:56,701
anecdotes, right? 
People talk about certain 

1124
00:54:56,701 --> 00:54:59,213
things, maybe through proxy, 
maybe even directly, right? 

1125
00:54:59,213 --> 00:55:01,443
Because they don't know the 
actual problems. 

1126
00:55:01,443 --> 00:55:04,162
And I think having a leaders 
that actually understand the 

1127
00:55:04,162 --> 00:55:06,758
details, right, dive deep is 
really crucial, especially in 

1128
00:55:06,758 --> 00:55:09,070
the technical team, right? 
No matter whether you are higher

1129
00:55:09,070 --> 00:55:10,498
up there or not. 
A hundred percent. 

1130
00:55:11,661 --> 00:55:14,569
So maybe I wanna set aside time 
as well for us to talk. 

1131
00:55:14,569 --> 00:55:18,205
Because you have been leaders 
within this Southeast Asian 

1132
00:55:18,205 --> 00:55:20,631
region. 
You know, you've been managing 

1133
00:55:20,631 --> 00:55:23,389
team in Indonesia, Singapore 
now, maybe Malaysia. 

1134
00:55:23,389 --> 00:55:25,790
Back to your, where you are 
from, right? 

1135
00:55:26,150 --> 00:55:28,625
And maybe a little bit of 
Vietnam and so many different 

1136
00:55:28,625 --> 00:55:30,125
countries. 
One unique thing about Southeast

1137
00:55:30,125 --> 00:55:33,176
Asia is all these different 
countries have very unique 

1138
00:55:33,176 --> 00:55:35,662
culture. 
Different languages as well, 

1139
00:55:35,662 --> 00:55:38,270
different working practices and 
things like that. 

1140
00:55:38,760 --> 00:55:42,095
And now coming to the, you know,
the advent of AI, right? 

1141
00:55:42,095 --> 00:55:46,103
So many people talk about AI. 
Do you think the southeast Asian

1142
00:55:46,103 --> 00:55:48,275
engineering talents are ready 
for this? 

1143
00:55:49,715 --> 00:55:53,760
Yeah, it's, uh, it's definitely 
a topic very close to my heart. 

1144
00:55:53,988 --> 00:55:57,348
As you know, and this is one of 
the reasons why I enjoy working 

1145
00:55:57,348 --> 00:56:00,168
at Grab, because Grab is totally
Southeast Asian first, and we 

1146
00:56:00,168 --> 00:56:03,748
are plugged into it and we have 
engineering teams across the 

1147
00:56:03,748 --> 00:56:08,121
region, which is exciting. 
So let's talk facts. 

1148
00:56:08,121 --> 00:56:15,311
I think from pure raw talent, 
capability, we are there, right?

1149
00:56:15,311 --> 00:56:17,731
Like if you think, we have the 
potential. 

1150
00:56:18,107 --> 00:56:22,283
You know, Southeast Asia, and 
they they just had the ASEAN 

1151
00:56:22,283 --> 00:56:24,820
Conference in Malaysia, two, 
three weeks back. 

1152
00:56:25,119 --> 00:56:28,929
Singapore PM was there. 
Our Malaysian PM was there. 

1153
00:56:29,439 --> 00:56:31,539
They talked about Southeast Asia
collaboration, right? 

1154
00:56:31,539 --> 00:56:33,019
And then they kinda shared some 
numbers. 

1155
00:56:33,019 --> 00:56:36,530
So, I think Southeast Asia in 
total is like 700 million 

1156
00:56:36,530 --> 00:56:39,531
people. 
Still pretty young, demographic 

1157
00:56:39,531 --> 00:56:42,297
wise. 
By total size of economy, I 

1158
00:56:42,297 --> 00:56:45,510
could be wrong here, but like 
maybe fifth or sixth in the 

1159
00:56:45,510 --> 00:56:48,091
world. 
The trick though is it needs to 

1160
00:56:48,091 --> 00:56:53,066
be seen as a whole. 
The strength is when we can 

1161
00:56:53,066 --> 00:56:57,356
figure out how to collaborate as
countries, leveraging our 

1162
00:56:57,356 --> 00:57:01,961
different strengths, right? 
If you are taken apart 

1163
00:57:01,961 --> 00:57:05,666
individually, all bets are off. 
That's my sense. 

1164
00:57:05,726 --> 00:57:08,111
Even Singapore. 
Singapore I think is probably 

1165
00:57:08,111 --> 00:57:10,161
definitely on the forefront. 
I mean, that's part of the 

1166
00:57:10,161 --> 00:57:11,386
reason why we are both here, 
right? 

1167
00:57:12,095 --> 00:57:15,034
I think the quality of the 
institutions, the schools are 

1168
00:57:15,034 --> 00:57:16,930
there, right? 
Like the amount of folks I, I'm 

1169
00:57:16,930 --> 00:57:19,520
sure if you go and look at like,
you know, research papers and 

1170
00:57:19,520 --> 00:57:22,640
you kind of look at like, where 
you'll find lots of folks who 

1171
00:57:22,640 --> 00:57:27,564
did their either undergraduate 
or masters or or PhD in 

1172
00:57:27,564 --> 00:57:30,552
Singaporean institutions, and 
you'll find folks who are from 

1173
00:57:30,552 --> 00:57:31,948
all across Southeast Asia as 
well, right? 

1174
00:57:31,948 --> 00:57:35,053
Like, you have lots of 
Malaysians doing stuff here, 

1175
00:57:35,053 --> 00:57:40,060
also Indonesians, right? 
It'll be great if we can build 

1176
00:57:40,060 --> 00:57:43,996
more local entities that are 
really harnessing all these 

1177
00:57:43,996 --> 00:57:47,416
different talent and applying 
them on more complex problems. 

1178
00:57:47,686 --> 00:57:49,711
Of course, Grab has a role to 
play here. 

1179
00:57:50,266 --> 00:57:53,236
I won't speak on behalf of Grab,
but I think we have definitely 

1180
00:57:53,236 --> 00:57:57,924
been trying to do more of that. 
You know, other large regional 

1181
00:57:57,924 --> 00:58:01,280
companies like S-E-A, Sea, I 
think is also doing some 

1182
00:58:01,280 --> 00:58:04,359
interesting stuff. 
I don't know, have you seen any 

1183
00:58:04,359 --> 00:58:06,012
other regional company doing 
some interesting stuff? 

1184
00:58:06,492 --> 00:58:09,616
I mean back then when there used
to be a lot of investment coming

1185
00:58:09,616 --> 00:58:12,707
to Southeast Asia, you know, 
tech startups are booming, I can

1186
00:58:12,707 --> 00:58:15,822
see a lot of investments. 
But these days, you know, the 

1187
00:58:15,822 --> 00:58:19,582
situation is quite tough, right?
And everyone having layoff. 

1188
00:58:19,952 --> 00:58:22,422
Uh, maybe outsourcing to 
different countries for, I don't

1189
00:58:22,422 --> 00:58:24,828
know, for whatever reasons. 
Maybe it's talent, maybe it's 

1190
00:58:24,828 --> 00:58:29,031
pay or whatever that is, right? 
And I just see us in Southeast 

1191
00:58:29,031 --> 00:58:31,592
Asia is a bit tough in terms of 
situation. 

1192
00:58:31,592 --> 00:58:33,545
And when you mention about 
collaboration, I don't know 

1193
00:58:33,545 --> 00:58:36,484
whether it's happening now. 
I would say even people are 

1194
00:58:36,484 --> 00:58:38,554
putting them back into silos, 
right? 

1195
00:58:38,764 --> 00:58:41,204
Maybe focusing more on the 
country rather than 

1196
00:58:41,204 --> 00:58:44,529
collaborating with each other. 
And what can we do as a tech 

1197
00:58:44,529 --> 00:58:45,810
community? 
Probably, that's one question, 

1198
00:58:45,810 --> 00:58:48,855
which is also sometimes I think 
about Tech Lead Journal, right, 

1199
00:58:48,855 --> 00:58:52,129
how can I serve this mission of,
you know, building upskilling 

1200
00:58:52,129 --> 00:58:54,019
the tech talents within the 
region, right? 

1201
00:58:54,229 --> 00:58:57,843
So maybe in your view how can we
approach this as a tech 

1202
00:58:57,843 --> 00:58:58,789
community, right? 
Yeah. 

1203
00:58:59,059 --> 00:59:01,599
So I think that regional tech 
players have a role to play. 

1204
00:59:01,739 --> 00:59:06,247
So like the Grab and Sea, right?
Like, we really need to be 

1205
00:59:06,247 --> 00:59:10,519
looking at how can we create 
opportunities for, you know, 

1206
00:59:10,519 --> 00:59:15,539
really high potential talents to
come and join and solve deep 

1207
00:59:15,539 --> 00:59:18,621
technical problems. 
From my experience kind of at 

1208
00:59:18,621 --> 00:59:22,429
Grab, like I think like the 
quality pool is there, coming 

1209
00:59:22,429 --> 00:59:24,555
out the door, coming out the 
gate. 

1210
00:59:24,735 --> 00:59:26,955
We have high potential 
individuals. 

1211
00:59:27,345 --> 00:59:30,033
It's just subsequent to that, 
can we give them the right 

1212
00:59:30,033 --> 00:59:32,205
opportunities to continue to 
grow and stay within the region.

1213
00:59:32,773 --> 00:59:35,053
I mean, I think government has a
role to play here, right? 

1214
00:59:35,053 --> 00:59:38,177
Like how do they set up the 
right type of investment 

1215
00:59:38,177 --> 00:59:40,930
structure so that companies, 
regional companies can kind of 

1216
00:59:40,930 --> 00:59:44,365
invest and build talent in other
countries, but also kind of 

1217
00:59:44,365 --> 00:59:47,273
somehow accrue it back to the 
country that's supporting that 

1218
00:59:47,273 --> 00:59:49,397
investment? 
So some type of like 

1219
00:59:49,397 --> 00:59:51,680
multilateral Asian framework 
needs to come out of this. 

1220
00:59:51,680 --> 00:59:54,050
I don't think we're gonna be 
able to solve this individually.

1221
00:59:54,593 --> 00:59:57,473
And then I think the regional 
tech companies have a role to 

1222
00:59:57,473 --> 01:00:01,078
play. 
Lastly, I think also the 

1223
01:00:01,078 --> 01:00:04,812
schools, the institutes have a 
very important role to play. 

1224
01:00:04,812 --> 01:00:09,212
Like, you know, we need to be 
better plugged into the 

1225
01:00:09,212 --> 01:00:11,096
industry. 
I think things are improving. 

1226
01:00:11,096 --> 01:00:15,224
Like I look at the internship 
programs that we have, Grab has 

1227
01:00:15,224 --> 01:00:19,025
with, in Vietnam, in Malaysia, 
and even in indonesia. 

1228
01:00:19,085 --> 01:00:23,987
And it's much better than what 
it was maybe in 2012, 2013 when 

1229
01:00:23,987 --> 01:00:25,657
I was also looking at it 
closely. 

1230
01:00:26,404 --> 01:00:29,132
And giving people, so like for 
example, Waterloo University in 

1231
01:00:29,132 --> 01:00:31,318
Canada. 
It's common for a software 

1232
01:00:31,318 --> 01:00:34,440
engineer, a CS graduate there to
have like four, five 

1233
01:00:34,440 --> 01:00:37,426
internships, and they kind of 
learn so much through the 

1234
01:00:37,426 --> 01:00:38,904
process. 
Why can't we have something 

1235
01:00:38,904 --> 01:00:40,944
equivalent here? 
I think partly, it's the 

1236
01:00:40,944 --> 01:00:42,584
companies, like we need to have 
the mindset. 

1237
01:00:42,614 --> 01:00:44,592
Right mindset. 
But I think this is where the 

1238
01:00:44,592 --> 01:00:46,750
institutes can go to companies 
and say like, okay, this is what

1239
01:00:46,750 --> 01:00:48,724
we're hoping, these are the type
of programs we're hoping to 

1240
01:00:48,724 --> 01:00:51,344
create for our students. 
Are you guys willing to kinda 

1241
01:00:51,344 --> 01:00:53,588
engage at the same level, right?
No point in getting an intern 

1242
01:00:53,588 --> 01:00:55,374
who then comes in and like, 
this, and that. 

1243
01:00:55,633 --> 01:00:57,763
Like just pushes paper, that's 
pointless. 

1244
01:00:58,123 --> 01:01:01,483
So those are some things I think
that will need to change and 

1245
01:01:01,483 --> 01:01:04,493
this is where like some type of 
collaboration, I think, is gonna

1246
01:01:04,493 --> 01:01:06,799
be key. 
Again, I wanna emphasize a point

1247
01:01:06,799 --> 01:01:10,389
that like I have been impressed 
by the quality of people coming 

1248
01:01:10,389 --> 01:01:13,846
out the gate. 
Now it's really a question of 

1249
01:01:13,846 --> 01:01:15,604
how do we create the right 
opportunities. 

1250
01:01:17,014 --> 01:01:19,918
You also mentioned one thing 
very interesting right? 

1251
01:01:19,918 --> 01:01:22,820
Because I've been also thinking 
about this and trying to find 

1252
01:01:22,820 --> 01:01:26,096
observations, right? 
This region seems to lack a 

1253
01:01:26,096 --> 01:01:29,144
little bit of deep technical 
problems and challenges 

1254
01:01:29,144 --> 01:01:30,976
probably. 
And that's why maybe more 

1255
01:01:30,976 --> 01:01:34,089
people, you know, going out to 
other countries, US maybe, or 

1256
01:01:34,089 --> 01:01:38,589
even China these days. 
And even if you just observe, 

1257
01:01:38,589 --> 01:01:40,761
right? 
Like for example, AI, right? 

1258
01:01:40,821 --> 01:01:44,201
There are not many people 
building AI solutions within 

1259
01:01:44,201 --> 01:01:46,800
this region. 
There's a lack in developer 

1260
01:01:46,800 --> 01:01:50,121
tools kind of a product being 
built around here. 

1261
01:01:50,601 --> 01:01:52,221
And you mentioned talent is 
there actually. 

1262
01:01:52,721 --> 01:01:56,781
So what is the gap actually for 
us to actually, you know, bridge

1263
01:01:56,781 --> 01:02:00,089
to that stage where we want to 
deep, invest deep into these 

1264
01:02:00,089 --> 01:02:01,714
kind of technical challenges as 
well? 

1265
01:02:01,714 --> 01:02:04,908
Or is it just difficult for us 
to have this kind of 

1266
01:02:04,908 --> 01:02:07,356
opportunity? 
So first off, so I agree, it's 

1267
01:02:07,356 --> 01:02:09,706
definitely a problem, but it's 
not just a Southeast Asia 

1268
01:02:09,706 --> 01:02:12,304
problem. 
If you look at it like outside 

1269
01:02:12,304 --> 01:02:17,460
of like the China and US, maybe 
some small parts of Europe like 

1270
01:02:17,460 --> 01:02:22,712
Mistral in France, the AI like 
agglomeration of companies are 

1271
01:02:22,712 --> 01:02:25,774
very concentrated, right? 
So I think it's a, it's a more 

1272
01:02:25,774 --> 01:02:27,526
wider problem, but it's 
definitely pretty acute in 

1273
01:02:27,526 --> 01:02:30,804
Southeast Asia as well. 
Again, I think it comes down to 

1274
01:02:30,804 --> 01:02:33,130
our opportunities, right? 
And an investment, like 

1275
01:02:33,130 --> 01:02:37,270
investment to be able to do 
something like a foundational 

1276
01:02:37,270 --> 01:02:41,448
model is huge as everybody 
really knows. 

1277
01:02:41,448 --> 01:02:46,602
And that itself means that only 
like, I think, within the 

1278
01:02:46,602 --> 01:02:48,830
region, like it's only 
governments that can really fund

1279
01:02:48,830 --> 01:02:50,994
these type of things. 
Now the challenge then for 

1280
01:02:50,994 --> 01:02:53,940
government to kind of fund these
things is like, what's the ROI? 

1281
01:02:54,540 --> 01:02:56,598
And it's a good question to ask 
because there's no point just 

1282
01:02:56,598 --> 01:02:58,200
doing these things for the sake 
of doing them. 

1283
01:02:59,596 --> 01:03:02,226
That's one dimension. 
Where we've been very good at is

1284
01:03:02,226 --> 01:03:04,802
more on the consumer app side. 
Like what you're saying, right? 

1285
01:03:04,802 --> 01:03:08,156
So less on the deep tech side. 
But what is hopefully going to 

1286
01:03:08,156 --> 01:03:10,562
happen is there's gonna be a 
confluence. 

1287
01:03:10,892 --> 01:03:15,340
So as we work out what type of 
use cases AI is really good at 

1288
01:03:15,340 --> 01:03:19,148
and we bridge it back to solving
customer problems, these two 

1289
01:03:19,148 --> 01:03:22,964
worlds will converge again, 
where solving consumer problems 

1290
01:03:22,964 --> 01:03:29,184
via AI will just be the norm. 
Today, I think we're still at 

1291
01:03:29,184 --> 01:03:32,656
the early stage curve, where 
we're almost looking for 

1292
01:03:32,656 --> 01:03:36,852
problems to solve with AI. 
But we are shifting, right, like

1293
01:03:36,852 --> 01:03:39,712
of course, there's like the 
coding use case, which is pretty

1294
01:03:39,712 --> 01:03:41,502
clear. 
And then there are gonna be 

1295
01:03:41,502 --> 01:03:44,756
other use cases as well where...
and that will kind of land into 

1296
01:03:44,756 --> 01:03:47,196
more clear ROI and we'll be able
to get there. 

1297
01:03:48,010 --> 01:03:51,574
But it'll take a bit of time. 
And it will take us kind of 

1298
01:03:51,574 --> 01:03:53,878
fixing some of the fundamentals,
but that's the only way to do 

1299
01:03:53,878 --> 01:03:57,112
it. 
I think we need to merge this 

1300
01:03:57,112 --> 01:04:00,565
consumer app problem, or 
consumer space problem back with

1301
01:04:00,565 --> 01:04:02,886
the AI problem. 
The use cases need to become 

1302
01:04:02,886 --> 01:04:06,194
clearer. 
I mean, anecdotally when I see, 

1303
01:04:06,194 --> 01:04:09,427
you know, tech forums within 
this region, people are now 

1304
01:04:09,427 --> 01:04:11,264
questioning should I still be in
tech? 

1305
01:04:11,264 --> 01:04:13,244
You know, should I still study 
computer science? 

1306
01:04:13,484 --> 01:04:15,219
Should I now thinking about 
something else? 

1307
01:04:15,824 --> 01:04:18,554
And for tech leaders as well, 
should I now think of, I don't 

1308
01:04:18,554 --> 01:04:20,624
know, moving to some other 
industries? 

1309
01:04:20,834 --> 01:04:23,928
In your view, right, as a tech 
leader and also being hands on 

1310
01:04:23,928 --> 01:04:27,497
with this AI thing these days, 
do you think we should start 

1311
01:04:27,497 --> 01:04:30,134
thinking that way? 
Or do think we should even 

1312
01:04:30,134 --> 01:04:33,254
double down on this? 
It's a good question. 

1313
01:04:33,423 --> 01:04:37,084
I'm opinionated about this. 
And of course, this is my 

1314
01:04:37,084 --> 01:04:40,601
opinion. 
I'm not, so like, I think it 

1315
01:04:40,601 --> 01:04:43,886
depends on what motivates you. 
If you took up computer science 

1316
01:04:43,886 --> 01:04:47,582
or if you're in this space 
because you want a good salary, 

1317
01:04:47,582 --> 01:04:51,169
there's nothing wrong with that.
It is a completely human and 

1318
01:04:51,169 --> 01:04:54,196
logical thing to do. 
Then yes, you might question 

1319
01:04:54,196 --> 01:04:56,794
these things because I think 
salaries are gonna be depressed.

1320
01:04:56,884 --> 01:04:58,384
I think there's gonna be a 
correction. 

1321
01:04:58,517 --> 01:05:01,071
There's been over hiring, 
there's be more layoffs, there's

1322
01:05:01,071 --> 01:05:03,017
gonna be efficiency gains that 
come out of it. 

1323
01:05:03,654 --> 01:05:08,581
But if you are in the space 
because you enjoy what it means 

1324
01:05:08,581 --> 01:05:12,225
to build good software, you're 
in it for a craft, you care 

1325
01:05:12,225 --> 01:05:15,279
about what you do, you should 
keep doing it. 

1326
01:05:15,339 --> 01:05:17,868
Because if anything, those 
things are what matters even 

1327
01:05:17,868 --> 01:05:20,562
more now. 
And this has been true all this 

1328
01:05:20,562 --> 01:05:23,261
while. 
It is just that in the past, the

1329
01:05:23,261 --> 01:05:25,741
value was more on the generation
of code. 

1330
01:05:26,161 --> 01:05:29,441
Can you get this stuff done? 
So then someone who didn't 

1331
01:05:29,441 --> 01:05:32,351
really care about it, but did 
enough work to kind of cut the 

1332
01:05:32,351 --> 01:05:35,121
website, get it out, build 
something quick for an agency, 

1333
01:05:35,121 --> 01:05:39,265
that was still valued. 
What you're seeing is all of 

1334
01:05:39,265 --> 01:05:41,371
that values going almost to zero
now, right? 

1335
01:05:41,371 --> 01:05:44,247
We talked about zero shot. 
Like, you know, image 

1336
01:05:44,247 --> 01:05:48,233
generation, website generation. 
So now those people who are kind

1337
01:05:48,233 --> 01:05:51,608
of operating, I think, at that 
level they are feeling this 

1338
01:05:51,608 --> 01:05:54,499
existential threat. 
Maybe the answer to them is, 

1339
01:05:54,499 --> 01:05:56,943
yeah, if this is what you want 
to do in software engineering, 

1340
01:05:56,943 --> 01:06:00,696
then yeah, you, there might be 
better ways to make money. 

1341
01:06:01,456 --> 01:06:04,366
Now, of course, this is gonna be
displacing people. 

1342
01:06:04,366 --> 01:06:06,840
There's gonna be, you know, 
economic costs around retraining

1343
01:06:06,840 --> 01:06:09,266
and stuff like that. 
So I'm well aware of all those 

1344
01:06:09,266 --> 01:06:12,506
things and I don't think we 
should take this sort of things 

1345
01:06:12,506 --> 01:06:14,786
uh, willy-nilly. 
These are, they're real 

1346
01:06:14,786 --> 01:06:16,464
ramifications to some of these 
things. 

1347
01:06:16,464 --> 01:06:21,942
But personally, I think it's, 
it's, it's, that's probably the 

1348
01:06:21,942 --> 01:06:25,937
direction we have to go. 
So in other words, to answer 

1349
01:06:25,937 --> 01:06:29,161
your question more directly, I 
think if you care about software

1350
01:06:29,161 --> 01:06:31,437
engineering, you should double 
down on it. 

1351
01:06:31,437 --> 01:06:35,837
But if you're just looking for 
very efficient ways of accruing 

1352
01:06:35,837 --> 01:06:38,666
wealth or earning a paycheck, 
maybe there might be other ways 

1353
01:06:38,666 --> 01:06:42,135
to do it. 
And I think we have kind of like

1354
01:06:42,135 --> 01:06:43,187
discussed it earlier as well, 
right? 

1355
01:06:43,187 --> 01:06:45,377
I mean, the standards would 
definitely elevate, right? 

1356
01:06:45,557 --> 01:06:50,342
So for those of you who are just
doing, I don't know, mundane 

1357
01:06:50,362 --> 01:06:53,432
tasks, you know, generating code
through hand, by hand, right? 

1358
01:06:53,782 --> 01:06:56,796
So I think those skills probably
will be, you know, lesser 

1359
01:06:56,796 --> 01:06:59,079
valued. 
And I think I like the point 

1360
01:06:59,079 --> 01:07:00,983
that you mentioned that Kent 
Beck mentioning right? 

1361
01:07:01,236 --> 01:07:04,175
A good software engineer, the 
10%, right, the skills that you 

1362
01:07:04,175 --> 01:07:06,689
actually have will actually be 
amplified in this era. 

1363
01:07:07,049 --> 01:07:10,769
And I still think, I'm betting 
as well myself, that hopefully 

1364
01:07:10,769 --> 01:07:13,415
with all good talents in 
software engineering that we 

1365
01:07:13,415 --> 01:07:16,582
have, right? 
Those 10% can still be valued in

1366
01:07:16,582 --> 01:07:19,637
the market, right? 
And just double down on that 

1367
01:07:19,637 --> 01:07:22,298
aspect rather than the 90% 
percent which is probably the 

1368
01:07:22,298 --> 01:07:25,777
more mundane stuff. 
So I'm very well conscious about

1369
01:07:25,777 --> 01:07:28,779
the time, Mohan. 
So as a tradition in my podcast,

1370
01:07:28,779 --> 01:07:32,436
I only have one last question. 
I call this technical leadership

1371
01:07:32,436 --> 01:07:34,256
wisdom. 
Think of it just like advice 

1372
01:07:34,256 --> 01:07:37,998
that you want to give us, to the
listeners, right, as your last 

1373
01:07:37,998 --> 01:07:39,795
advice. 
So what would that be? 

1374
01:07:41,120 --> 01:07:44,663
I think for people starting out 
early in their career, go wide. 

1375
01:07:44,903 --> 01:07:46,673
Don't try to specialize too 
quickly. 

1376
01:07:46,944 --> 01:07:51,228
Be willing to pick up anything. 
Learn, look more for the 

1377
01:07:51,228 --> 01:07:55,368
opportunities to learn either by
taking on challenges that are 

1378
01:07:55,368 --> 01:07:58,452
gonna stretch you and grow you 
or by finding the right coaches 

1379
01:07:58,452 --> 01:08:01,696
or mentors to learn from. 
That's something I would say. 

1380
01:08:02,266 --> 01:08:05,912
I think for people maybe in the 
middle stage of their careers, 

1381
01:08:06,516 --> 01:08:10,336
this is a great time to become 
more hands-on, to learn stuff. 

1382
01:08:10,769 --> 01:08:14,620
I think, again, quoting, you 
know, venerable Kent Beck, he 

1383
01:08:14,620 --> 01:08:19,501
said this is the most fun he's 
been having in like a 50 year 

1384
01:08:19,501 --> 01:08:22,273
software development career. 
Can you imagine 50 years of 

1385
01:08:22,273 --> 01:08:23,185
career? 
That is amazing! 

1386
01:08:23,734 --> 01:08:26,345
So yeah, I would definitely 
recommend people get back. 

1387
01:08:26,645 --> 01:08:28,973
Get back to your roots. 
Find time on the weekends, if 

1388
01:08:28,973 --> 01:08:31,774
you can't be doing office hours.
Carve a better time out. 

1389
01:08:31,965 --> 01:08:34,981
Spend some time here and there. 
You know, as we discussed, these

1390
01:08:34,981 --> 01:08:38,915
tools are really good at giving 
you what dopamine hits, right? 

1391
01:08:38,915 --> 01:08:42,649
You get a really good high ROI 
so find the time to kind of go 

1392
01:08:42,649 --> 01:08:45,269
in deeper. 
So it's been a pleasure. 

1393
01:08:45,269 --> 01:08:47,219
So I think we learned a lot, 
right? 

1394
01:08:47,219 --> 01:08:50,788
And especially looking from, you
know, your perspective, right? 

1395
01:08:50,788 --> 01:08:54,756
Like what makes a great leader? 
How do you build a great 

1396
01:08:54,756 --> 01:08:56,965
engineering team? 
And we touched on also like how 

1397
01:08:56,965 --> 01:08:59,751
do we approach this AI thing, 
right, that is happening in the 

1398
01:08:59,751 --> 01:09:02,745
software engineering team. 
And going back also to the 

1399
01:09:02,745 --> 01:09:06,506
Southeast Asian talents, right. 
So I think, I hope all the 

1400
01:09:06,506 --> 01:09:08,499
listeners here can learn from 
this conversation. 

1401
01:09:08,499 --> 01:09:10,599
And yeah, thank you so much for 
being here today. 

1402
01:09:10,809 --> 01:09:13,269
Thank you so much, Henry. 
I really appreciate you, you 

1403
01:09:13,269 --> 01:09:16,013
know, enlisting me for your 
podcast and hopefully this was 

1404
01:09:16,013 --> 01:09:16,742
useful for folks.