Building WhatsApp with Jean Lee | CEO OS

tldr

Jean Lee was engineer #19 at WhatsApp, joining when the company had almost no formal processes. A team of ~30 engineers supported 450M+ users across 8 platforms using Erlang, with no code reviews, no scrum, no feature flags, and no managers — just high-trust engineers shipping directly to production. WhatsApp charged $1/year which covered all costs. The $19B Facebook acquisition changed the culture dramatically, and Jean eventually transitioned from IC to engineering manager at Facebook London. She now advises startups and advocates for small, empowered teams with strong foundations.

Key Takeaways

30 engineers, 450M users. WhatsApp proved that a tiny, elite team can scale to global impact. No scrum, no agile, no TDD — just trust and dogfooding.
Say no to 99% of feature requests. Jan Koum's default was "no." The product stayed simple because they prioritized a grandma in a remote town being able to use the app reliably.
$1/year covered everything. Server costs, salaries, SMS verification — all covered by the annual fee. No ads, no investors pushing growth hacks.
Erlang was the secret weapon. Chosen because it was built for telecom-grade reliability. Hot code reloading meant zero-downtime deployments.
No code reviews, no managers. Engineers were trusted to push to production. The only "review" was your first commit. After that, you were trusted.
Dogfooding was the QA. Before every release, the team would use the new build internally. If something felt off, it didn't ship.
The $19B acquisition was announced in person — and one engineer missed it because she was getting her eyebrows done.
Facebook's culture was the opposite. Visibility mattered. You had to post your work internally to get recognized in perf reviews. A very different game from WhatsApp's heads-down culture.
Career advice: Focus on foundations. Languages and tools come and go. Systems thinking, problem-solving, and communication endure.
Empower people, don't micromanage. Both at WhatsApp and in her advisory work, Jean's philosophy is: hire smart people, give them context, get out of the way.

Detailed Breakdown

[00:00] Intro — The WhatsApp Teaser

Gergely opens with a montage of Jean's most striking claims: no code reviews, Jan saying no 99% of the time, no agile/scrum/TDD, and the $1/year model covering all costs. Sets up the episode as a deep dive into how one of the world's most-used apps was built by an absurdly small team.

[01:39] Early Years in Tech

Jean shares her path into engineering. She was job hunting and got a verbal offer from another company, but they were slow with the written offer. WhatsApp moved fast — Jan called her in, asked what it would take, and she signed the next day. She got the other company's written offer on her first day at WhatsApp. Lesson: Speed in hiring wins. The company that moves fastest gets the best people.

[06:18] Becoming Engineer #19 at WhatsApp

Jean describes the early WhatsApp office — tiny, scrappy, engineers with noise-cancelling headphones grinding away. The interview was surprisingly informal: a chat with Jan and Brian, a coding exercise, and that was it. No multi-round gauntlet. They were looking for strong engineers who could be trusted, not people who could pass algorithm puzzles.

[13:53] WhatsApp's Tech Stack

Erlang was the backbone. Why? Because it was designed for telecom systems — built for reliability, concurrency, and zero-downtime deployments via hot code reloading. This let WhatsApp push updates without taking the service offline. The team supported 8 platforms (iOS, Android, BlackBerry, Nokia S40, Nokia S60, Windows Phone, web, and more). Engineers owned entire platforms — Jean herself was the sole engineer for one of them.

[18:09] WhatsApp's Unique Ways of Working

This is the section that makes engineering managers twitch:

No code reviews (except your very first commit)
No managers — engineers reported directly to Jan and Brian
No scrum, no agile, no TDD
No feature flags, no canarying, no staged rollouts
Engineers pushed directly to production
The only safety net was aggressive dogfooding — every build was used internally before release
Communication happened in WhatsApp groups (naturally)
People read each other's git commits because there were only 30 engineers

Jean notes this worked because the team was small, elite, and deeply trusted. Everyone cared about quality. There was no process because the people were the process.

[25:27] Countdown Displays and Outages

WhatsApp had monitors showing real-time user counts and message volumes. When something went wrong, you could see it immediately. The team had a cultural obsession with uptime. Outages were rare but treated as all-hands emergencies. No incident retros or blameless postmortems — just fix it and make sure it doesn't happen again.

[27:07] Why WhatsApp Won

Jean's take: simplicity and reliability. While competitors were adding stickers, games, and social features, WhatsApp stayed laser-focused on messaging. It worked everywhere — on cheap phones, on slow networks, in countries with unreliable infrastructure. The $1/year model meant no ads, no data harvesting, no creepy business model. Users trusted it.

[28:53] The Facebook Acquisition ($19B)

The announcement came at an all-hands. Jan and Brian gathered everyone. One engineer was late — she was getting her eyebrows done. Jan and Brian were trying (and failing) to hide their excitement. When they announced the acquisition, Jean zoned out for a moment processing it. The number was surreal.

Jan had previously said he would never sell — that selling your company is like selling your baby. But the Facebook deal came with promises of independence, no ads, and continued focus on privacy.

[33:13] Life After Acquisition

The immediate aftermath was relatively smooth — WhatsApp was allowed to operate independently. But over time, Facebook's culture seeped in. More process, more meetings, more politics. The team that had operated on pure trust and speed started feeling the weight of a large organization. Some early employees left. The founders eventually departed over disagreements about ads and data sharing.

[39:27] Working at Facebook in London

Jean was one of the few WhatsApp engineers who relocated to London to help grow the office there. Facebook already had a presence, so logistics were handled. But building a new engineering hub meant hiring, establishing culture, and navigating the very different Facebook way of working.

[44:07] Transitioning to Management

Jean moved from IC (individual contributor) to engineering manager at Facebook. The transition was eye-opening. At WhatsApp, there were no managers. At Facebook, management was a craft with its own expectations, tooling, and performance criteria.

She describes the Facebook performance review system: engineers needed to make their work visible by posting about it internally. Managers who prepared thoroughly for calibration meetings — with data, examples, and narratives — could better advocate for their reports. Those who didn't prepare put their engineers at a disadvantage.

[47:27] Performance Reviews as a Manager

Jean's approach: she prepared obsessively for perf calibration. She'd gather evidence, write narratives, and anticipate questions. This gave her an edge in representing her engineers fairly. She notes this was somewhat unfair — your performance rating partly depended on how good your manager was at internal advocacy. AI might eventually level this playing field by automating the grunt work of evidence gathering.

[53:29] After Facebook

Jean left Facebook and moved into advisory and consulting work. She advises startups, particularly on engineering culture, hiring, and scaling small teams. Her experience at WhatsApp (tiny team, massive scale) and Facebook (big company, process-heavy) gives her a unique lens on both worlds.

[58:53] AI's Impact on Engineering

Jean's take on AI in software engineering:

AI will handle tedious, repetitive work (adding comments, boilerplate, simple features)
The engineers who thrive will be the ones with strong foundations — systems thinking, problem-solving, communication
Tools and languages are temporary. The ability to reason about complex systems is permanent.
She's not worried about AI replacing engineers. She's excited about AI removing the boring parts so engineers can focus on the interesting problems.

[1:02:34] Jean's Advice to New Grads and Startups

Focus on foundations. Data structures, algorithms, systems design — these don't go out of style.
Join a small team if you can. You'll learn 10x faster when you own entire systems.
Don't chase trends. The hot framework of 2026 will be legacy by 2030. Understanding why things work matters more than knowing how to use the tool du jour.
For startups: Stay small as long as you can. Every hire is a trade-off. WhatsApp proved 30 engineers can serve half a billion users.

[1:06:45] Empowering Employees

Jean's management philosophy: hire smart people, give them context and autonomy, then get out of the way. This was the WhatsApp model and it's what she advocates for in her advisory work. Micromanagement is a sign that you either hired wrong or you're not giving enough context.

[1:08:17] Book Recommendations

Jean shares her reading list (titles mentioned in the conversation for those interested in leadership, engineering culture, and building high-performing teams).

Notable Quotes

"99% of the time he would say no. All the cool features were missing in my mind, but that was by design." — Jean Lee, on Jan Koum's product philosophy

"Selling your company is like selling your baby." — Jan Koum (as recalled by Jean)

"We didn't use any of it... I'm surprised to hear they thought they were shipping faster because of it." — Jean Lee, on Scrum/Agile/TDD

"That $1 was enough to pay for the server cost, the salaries, and the SMS code every year." — Jean Lee, on WhatsApp's business model

"Foundations don't go anywhere. Tools come and go, languages come and go." — Jean Lee, career advice

One Thing to Act On

Say no more. WhatsApp served 450M users by rejecting 99% of feature requests. Next time a feature feels "nice to have" in FeatureOS or SupportWire — kill it. The product you don't build is the one that keeps the product you did build excellent.

Tags

#engineering #whatsapp #small-teams #scaling #hiring #product-philosophy #erlang #startup-culture #management #facebook #acquisition #pragmatic-engineer

Raw Transcript

Auto-captions from YouTube. Folded by default — expand if you need to grep the source or pull an exact quote.

0:00 I have a feeling WhatsApp was not 0:01 exactly a standard startup. 0:02 >> So, we didn't have code reviews, but the 0:05 only time I got my code reviewed was the 0:08 first time I made a commit. 0:09 >> And you said that Jan said no a lot. 0:12 >> 99% of the time he would say no. All the 0:15 cool features were missing in my mind, 0:18 but that was by design because we really 0:20 wanted to prioritize again the quality 0:22 of a grandma in a remote town being able 0:24 to use our app at any given time. 0:27 >> Scrum agile with a capital A TDD. Did 0:29 you use any of these at WhatsApp? 0:31 >> I'm surprised to hear they thought they 0:32 were shipping faster because of it. We 0:34 didn't use any of it. 0:35 >> So, you were break even. 0:36 >> Yeah, that $1 was enough to pay for the 0:40 server cost, the salaries, and the SMS 0:42 code every year. 0:47 >> Gene Lee was number 19 at WhatsApp. She 0:50 joined when hardly anyone in the US had 0:52 heard of it, saw it grow to 450 million 0:55 users, and was sitting at her desk with 0:57 noise cancelling headphones on when news 0:59 broke that Facebook bought them for $19 1:01 billion. In today's conversation, we 1:04 discuss how WhatsApp built natively 1:06 eight different platforms with a team of 1:08 30 engineers. Why the founders said no 1:10 to almost every feature request for 1:12 years. How WhatsApp's team operated with 1:14 no code reviews, no stand-ups, no sprint 1:17 planning, and many more. If you want to 1:19 understand how a tiny team with almost 1:21 no process built one of the most 1:22 successful products in history and what 1:24 today's AI native startups can still 1:26 learn from them, this episode is for 1:28 you. This episode is presented by 1:29 Statsig, the unified platform for flags, 1:31 analytics, experiments, and more. Check 1:34 out the show notes to learn more about 1:35 them and our other season sponsors, 1:37 Sonar and Work OS. Gan, welcome to the 1:40 podcast. It is amazing to to meet you. 1:43 You have quite the story early engineer 1:46 at at WhatsApp. But before we get into 1:48 WhatsApp, how did you get into tech? 1:51 >> I've always been a small town girl. My 1:52 dad was an OG hipster. He was really 1:54 into brewing beer. So, he decided to get 1:56 a PhD in beer. 1:58 >> In beer? 1:59 >> Yeah. In brewing. In brewing. Yeah. 2:02 >> So, I moved to San Francisco in 1999 and 2:04 that's when I got really exposed to all 2:06 the different tech roles. Like, growing 2:08 up, I didn't really even think about 2:11 engineering as a job. Um, of course I 2:14 use computers and I thought it was 2:15 really cool to be able to use Yahoo and 2:17 search things online, but beyond that, 2:20 uh, my first exposure to Silicon Valley 2:22 and tech came from living here. I got to 2:25 meet a lot of people who work in tech. I 2:27 dabbled around with coding when I was a 2:29 teenager, but not too seriously. But I 2:32 did think it was really cool that you 2:33 can just write a few lines and it will 2:36 just do things for you over and over and 2:38 over. It was almost magical. I I love 2:40 the feeling of creating something that 2:42 that actually runs um and debugging 2:44 something and fixing it and it runs 2:46 again. That that was really joyous and I 2:50 didn't really get into like super into 2:52 coding until I went to college. But one 2:54 of the reasons why I decided I wanted to 2:56 go into coding was I talked to different 2:59 people. So I thought maybe I want to be 3:01 a designer, maybe I want to be an 3:02 architect, maybe I want to be an 3:03 engineer. And I talked to different 3:05 adults who work in the in the industry. 3:07 After talking to a lot of adults, I 3:09 realized people who are in tech were the 3:11 only ones who were really excited about 3:13 their jobs. So in Silicon Valley, when 3:16 you ask people like tell me about your 3:18 work, people are often very hopeful for 3:20 the future and very proud of what 3:22 they're building. Compared to many other 3:25 adults that I spoke with, they were not 3:27 so encouraging. They're like, "Oh, don't 3:29 become an architect. Don't become a 3:31 designer." So that that was one of the 3:33 influences um for me early on. I studied 3:36 computer science at USC. Um, one of my 3:38 first internships, actual like coding 3:41 internships was at a small company. It 3:44 was a threeperson startup started by one 3:47 of the new grads from USC. And you'll 3:50 probably uh understand it was a video 3:52 sharing website, 3:55 but it was not like YouTube, but there 3:58 were so many versions of YouTube back in 4:00 the days before what YouTube was 4:02 dominant, right? So you probably 4:03 remember dozens of these like video 4:05 sharing platforms everywhere and one of 4:07 the issues of having so many options is 4:09 that you have to be visiting 12 4:11 different sites to search for new 4:12 things. So we had a website where you 4:15 can aggregate all the different types of 4:17 videos from different sources which is 4:19 actually kind of funny because lately 4:21 I've been seeing a lot of AI platforms 4:23 where you can just switch between the 4:25 models very similar to that. 4:27 >> Yeah. How did you get into IBM? I really 4:30 loved working for a small um threeperson 4:33 startup because I got to work with 4:37 engineers um we had engineers overseas 4:39 in China so I got to work with them. I 4:42 got to also do a little bit of coding 4:43 myself, but I was coming up with the 4:46 design docs like the the features list 4:48 and I was calling a lot of the shots and 4:50 I could also directly see the impact of 4:53 my code immediately on the website and I 4:55 thought that type of ownership and speed 4:58 and the visibility was really exciting 5:00 that I get to see the the impact of my 5:03 work immediately. 5:05 But one thing I wish I had was a little 5:08 bit more mentorship because we were all 5:10 new grads and in college um I felt like 5:13 we were just shooting things to see 5:15 which sticks. Um and I thought maybe for 5:18 my first job out of school I would like 5:20 a little bit more mentorship and 5:22 training and I started looking at more 5:24 bigger companies more traditional 5:26 companies and that's how I ended up at 5:28 at the time it was literally the biggest 5:30 company in the US. At what point did you 5:33 decide that you wanted to leave or try 5:35 out something else or did you even 5:36 decide or something just came up? 5:37 >> One of the reasons why I wanted to go to 5:39 a more traditional company with more 5:41 structure was so that I could get more 5:42 mentorship and training and IBM was 5:45 excellent for that. There were so many 5:47 veterans. They had so much experience 5:49 and they were willing to share with me 5:50 because they were 20 30 years ahead of 5:53 me, right? But one thing I really missed 5:56 was the small team environment. It was 5:59 just so big. There was a lot of 6:02 meetings, a lot of process and I I 6:04 missed seeing the impact of my work. I 6:06 couldn't quite understand how my work 6:09 was contributing to the overall company. 6:11 So then I decided to take some time off 6:14 and explore and have some fun. 6:16 >> Yeah. And around what time was this? 6:18 What year was this? 6:19 >> So I started working 2007 and I left by 6:22 2009 which was actually in retrospect I 6:25 was really brave because it was in the 6:27 midst of economic downturn. My thought 6:29 process at the time was I was only 22 or 6:33 three and I figure even if I take a year 6:35 off I can still catch up which I did. 6:38 >> And what what happened from there? How 6:40 did you eventually get to WhatsApp? That 6:42 was years later right? 6:44 >> Yeah. So I took some time off to try out 6:48 different like classes. I took a lot of 6:50 classes. I did a little bit of now 6:53 nowadays you call it the gig work but I 6:55 did all kinds of work. So whatever I 6:57 needed to you know make a living um 7:00 while taking all these classes and 7:03 exploring and really finding out what 7:05 like what kind of environment or what 7:07 kind of career do I envision for myself 7:09 and after I took those time off I 7:12 decided that I want to go back to 7:13 Silicon Valley but this time I do want 7:16 to work for a startup but maybe with 7:19 people who are a little bit more 7:20 experienced maybe not new grads and 7:22 maybe not a threeperson startup but a 7:24 little bit more stable startup where I 7:26 can possibly get both the the autonomy 7:29 and the the impact of the work but also 7:33 a little bit more mentoring because I 7:35 was still in my 20s. 7:36 >> Okay. So, how did you find this startup 7:38 which of course happened to be WhatsApp 7:40 >> in 2012? WhatsApp was still early. They 7:44 started in 2009 and they did still have 7:47 a lot of users but they're mostly in 7:49 Europe and in India. Um they were not 7:52 very known in America. Were you a 7:55 WhatsApp user back then? 7:56 >> I was not, but my my wife and her 7:58 friends were or or back then my you know 8:00 my my my girlfriend. But so some of my 8:02 friends were using it on and off. It was 8:03 kind of starting to be big in Europe. It 8:06 wasn't as massive just yet. 8:07 >> Exactly. Um I was lucky because I I 8:09 actually lived in New York for a little 8:11 bit before moving here and a lot of 8:14 people in New York were using it because 8:15 it's an international hub. So I I had 8:18 used the pro product in the past and I 8:20 saw the job posting on LinkedIn 8:23 >> and then you applied. What was the 8:25 interview like? 8:26 >> I don't think we did any leak code until 8:28 way way later until when we started 8:31 hiring interns and new grads. Most of 8:34 the interviews were talking about I I 8:38 guess you can call it system design 8:40 interviews. We would talk about how 8:42 would you design this, how would you 8:44 design that, like tell me about your 8:46 past experience building this product 8:48 and I recall talking to Yan about 8:52 different messaging apps and being 8:54 Korean, I told him a lot about Cacao 8:56 Talk and how it worked. Yeah, that was 8:58 my interview. 8:59 >> Just like that, you you got an offer. I 9:01 guess it's startup, right? Things move 9:02 fast like I assume it must have been 9:05 quick turnaround offer and then you had 9:07 to decide, right? How did you decide 9:08 that you're going to join this 9:10 relatively unknown startup that is 9:12 building some cool messaging that you 9:14 kind of thought was cool? But there 9:16 wasn't much information about that. In 9:17 fact, their glass door rating at the 9:19 time I remember had a one star. It had 9:21 one review, one star, someone saying, 9:23 "Oh, I don't like working here or who 9:25 knows if that was even a real employee, 9:27 but that was their glass door." 9:28 >> Oh, that's so interesting. I don't 9:30 remember looking up. I must have looked 9:32 up glass door, but like I was really 9:34 lucky because I actually had another 9:36 offer from a different company, but they 9:39 were a little bit sore. 9:41 >> One company was taking weeks to get Gene 9:42 an offer letter. Another founder closed 9:44 the deal in person the very next day. 9:47 Speed matters and not just in hiring. 9:50 This leads us nicely to our season 9:51 sponsor, work OS. AI startups are 9:53 reaching enterprise customers faster 9:55 than ever, sometimes just months after 9:57 launch. And in the moment that happens, 9:59 the requirements change. Customers wants 10:01 SSO, SCIM, audit logs, and granular 10:04 permissions before they'll deploy. 10:06 Building that infrastructure yourself 10:07 takes months. Works gives you APIs to 10:09 ship it in days. Authentication, SSO, 10:12 SCIM, Arbback, audit logs, and more. All 10:15 designed to integrate directly into your 10:17 product. Skip the rebuild, keep 10:19 shipping, visit work.com. And now, let's 10:21 get back to Gan and how the other 10:23 company could not get her written offer 10:25 as quickly as WhatsApp did. It was not a 10:27 startup and they said, "Oh, hey, like 10:29 you have my verbal offer. I am going to 10:31 give you a written offer soon." But then 10:34 it took them a while and meanwhile 10:38 um Yan called me few days later after 10:41 the interview and he said come into the 10:43 office right like today or tomorrow. 10:46 >> Yeah. 10:47 >> And then he asked me what would it take 10:49 for you to take the offer right now. 10:52 >> Love it. What did you say? I mean I 10:55 wasn't looking for that much. I mean I 10:57 was in my 20s. So I just told them oh 10:59 like few things I would like to have 11:00 then sure I'll take the offer and I saw 11:03 signed the offer the following day. And 11:05 I did actually hear back from the other 11:08 company on the first day I started 11:10 WhatsApp. They called me and I was like 11:12 oh I just started a new company. 11:15 >> That's it with start of you move faster 11:17 otherwise don't be surprised. So you 11:19 were engineer or you were employee 11:21 number 19 at WhatsApp right? was 11:23 engineer number 19. 11:24 >> Engineer number 19 at at at WhatsApp. 11:26 And you told me something really 11:28 interesting that you were the youngest 11:29 person even though you were like by this 11:31 time at your mid mid20s or or so. 11:33 >> I thought about that. So I recall there 11:36 were about four of us under the age of 11:39 30. So I was not the young guest, but 11:42 there were two people who were new grads 11:45 and then myself and one other person who 11:48 were in our late 20s. But the other like 11:51 15 or so people above 30 at a startup 11:53 which is kind of unheard. Why do you 11:56 think this was? This is so interesting. 11:58 >> That is true. Is it still rare nowadays? 12:01 Like 12:03 >> good question. I I think these days it 12:05 might not be as rare by the way. 12:07 >> I think so because I think I read some 12:09 kind of statistics from investors that 12:12 actually when they look at the success 12:14 rates of startups they found that older 12:17 uh founders tend to do better. Yeah. And 12:20 and then WhatsApp I guess you know like 12:22 Jana and and Brian they they started 12:24 this at like mid30s or or so after they 12:27 spent like more than a decade working at 12:29 Yahoo and other places. 12:30 >> Exactly. 12:31 >> Yeah. So I guess they must have been 12:33 able to hire like their network whatnot. 12:35 >> Yeah. The first 10 or so engineers a lot 12:38 of them came from Yahoo. Um some came 12:41 from Europe. You mentioned the story 12:43 when Yan reached out to you. Um Yan used 12:46 to do that. you would just look up who 12:47 is the expert in this field and reach 12:50 out to people and we had a lot of 12:51 contractors in Europe and then we had 12:54 some like mostly from personal 12:56 connection like from Stanford because 12:58 Brian went to Stanford and then we had 13:00 some referrals from Sequoia because they 13:02 invested in WhatsApp. It is fascinating 13:05 because the way we connected actually is 13:07 is both of us know Yan. I mean you've 13:09 worked with him but I I had an inmail in 13:11 my inbox from him I think six months 13:13 before you joined WhatsApp where I got a 13:15 message from him and saying hey I I 13:17 built a Windows phone app at the time 13:19 together with my brother called Cocktail 13:21 Flow and it was a beautiful Windows 13:23 phone app and it was labeled career 13:24 opportunity. So what you're saying is 13:26 there's a alternative timeline where if 13:29 I said like yes I'm interested which in 13:31 hindsight if a founder reaches out you 13:33 probably should at least talk to them 13:34 don't make the mistake that I did which 13:36 is just saying like I'm sorry I'm busy 13:38 if I might have been a contractor from 13:39 Europe so like sounds like that that was 13:41 a strategy and that was a smart 13:42 strategy. 13:43 >> Yeah we had many contractors in Europe 13:45 and they were all very experienced 13:47 people they were basically managing 13:49 themselves. We had people all over the 13:51 world working with us. What was a tech 13:53 stack like at WhatsApp? Before Gene 13:56 walks us through one of the most unusual 13:58 tech stacks in startup history, we're 14:00 talking about eight platforms and a 14:01 handful of engineers, let's talk about 14:03 keeping your code base healthy with our 14:05 season sponsor, Sonar. Sonar, the makers 14:08 of Sonar Cube, helps you automate code 14:10 review and verify the quality and 14:12 security of your code across your entire 14:14 stack so bugs don't make it to 14:16 production, whether you've got one 14:17 platform or eight. As agents take over 14:20 the development process, Sonar has 14:22 introduced the agentcentric development 14:23 cycle framework AC/DC, a new software 14:26 development methodology designed for the 14:28 unique scale and speed of AI generated 14:30 code. It's a move towards a more 14:32 intentional four-stage loop that gives 14:34 agents the guardrails they actually 14:36 need. The four phases being guide. 14:40 First, agents need to understand the 14:42 canvas on which they're being asked to 14:44 create so that the output fits with what 14:46 the developer and organization require. 14:48 Generate the LMB based tool generates 14:50 the code it believes will achieve the 14:52 desired outcome within the right 14:53 context. Verify. Next, the agent is 14:57 deliberately required to check its work, 14:59 ensuring that it actually achieves the 15:01 desired outcomes and is reliable, 15:03 maintainable, and secure. Solve. 15:06 Finally, any issues identified are 15:08 provided to a code repair agent to fix. 15:10 To power this, Sonar has significantly 15:12 strengthened its offering, introducing 15:14 products and capabilities like Sonar 15:16 context augmentation, Sonar Cubagentic 15:18 analysis, Sonar Cube architecture, and 15:21 Sonar Cube remediation agent. Head to 15:23 sonarsource.com/pragmatic 15:25 to learn more about the latest with 15:26 sonar and how it's empowering 15:28 organizations to embrace the agentic 15:29 era. And with this, let's get back to 15:31 Gene and all the different tech stacks 15:32 that WhatsApp had. 15:34 >> We were actually pretty unique. I I 15:36 don't think any startup ever really does 15:38 this, but we had seven different stacks. 15:41 We had I actually looked it up because 15:44 it's hard to count them all. We had, of 15:46 course, everybody has iPhone and 15:47 Android, but we also had Blackberry and 15:50 Windows Phone, which is also pretty 15:52 common, but we also had uh Nokia S40, 15:55 S60. We had a thing called Kyios for a 15:59 while, but not for a long time. And we 16:01 had the web client, so it's actually 16:03 eight. So, so you have of course you 16:05 know we know that iOS is Objective C, 16:08 Android was Java back in the day and 16:10 that all of these like the the 16:11 Blackberry the Nokia they all had I 16:13 think Nokia was Symbian C++ they all had 16:15 like their own different language and 16:18 then we've not talked about the back end 16:19 right 16:19 >> and the back end was Erlang 16:21 >> Erlang can you tell us about Erlang 16:24 because this that is one of the most 16:26 exotic tech sack I've heard Erlang in 16:28 telecommunications context at Ericson 16:31 again in Europe it is popular with the 16:33 telos, but startup wise I'm not sure I 16:36 heard anyone else use Erlang. 16:38 >> You might be right. They do have a 16:40 Erlang conference. I think it's called 16:42 Erling Factory. There's a really great 16:44 talk by one of our engineers, Rig Reed, 16:47 if you're interested in learning more 16:48 about it, but 16:49 >> we'll link it in the show notes below. 16:51 >> I'm I'm pretty sure it's still on 16:52 YouTube. I haven't looked up recently, 16:54 but uh he gave a really great talk about 16:56 why they started working with Erling and 16:58 it was the perfect choice. And he 17:00 describes it as um trying to maintain 17:04 the engine of an airplane while it's 17:07 flying 24/7 17:09 because if you imagine like WhatsApp is 17:11 so international, we can't take a break, 17:13 right? We have to continuously keep 17:15 running and it's always busy. Someone's 17:18 it's 8:00 a.m. somewhere in the world, 17:20 right? and Erling was a really robust um 17:24 language that was really good at 17:26 concurrencies and they stumbled upon it 17:29 because they were using this other tool 17:30 that happened to use Erling and decided 17:33 this is the perfect language 17:34 >> and I guess at the core of WhatsApp what 17:37 was the core engineering challenge was 17:38 it like so many messages being kind of 17:41 coming in needing to be seated out and 17:43 sent to different you know platforms 17:45 >> yeah that was one of the main challenges 17:47 like for example for New Year's or 17:49 Christmas because everyone's saying 17:52 happy new year at the exact same moment. 17:54 That was always our big uh biggest 17:56 challenges every year and we would 17:58 celebrate hey we didn't we didn't go 18:00 down after New Year's. So the the 18:02 interesting thing about the seven 18:04 different mobile platforms specifically 18:06 is the conventional wisdom wisdom before 18:08 and after has been like look if you want 18:10 to support all those platforms don't be 18:12 silly do crossplatform either build your 18:15 own layer that is crossplatform or use 18:18 you know there's all sorts of frameworks 18:20 why did WhatsApp not do this do you 18:21 remember the discussions of like why why 18:24 hire seven including some really hard to 18:26 hire people like for Nokia and Symbian 18:28 and you mentioned contractors in Europe 18:30 I mean sounds a bit of a nightmare Why? 18:32 >> So, Yan used to always say, "I want a 18:36 grandma in a remote countryside to be 18:38 able to use our app." So, what does that 18:40 mean? They may not have the newest 18:43 iPhone, the shiniest phone with the 18:45 biggest memory, right? In the 18:47 countryside where a grandma is using it, 18:49 you need the app to be lightweight. You 18:52 need it to work on any kind of device, 18:55 and you need the app to be simple. So 18:58 those were our um goals and priorities 19:01 and uh that's the thought process that 19:03 went into our decision to build seven 19:06 different platforms 19:07 >> and then inside WhatsApp how did you get 19:09 things done? Do you remember like how a 19:12 project got done or what was the concept 19:13 of projects and kind of what engineering 19:15 processes people might have followed 19:17 especially you know later you worked at 19:18 at meta compared to like how you know 19:21 like more kind of you know standard 19:22 startups work because I have a feeling 19:24 WhatsApp was not exactly a standard 19:26 startup was it? 19:28 >> Not really. Um, even meta compared to 19:32 other big tag, especially when I was at 19:35 Meta, was pretty scrappy, like not so 19:37 much on writing documents, for example, 19:40 the the move fast and break things motto 19:43 kind of allowed them to be a little bit 19:46 more lean in terms of their process. Um, 19:49 at least while I was there, but was like 19:52 the ultimate lean company. By the time 19:55 we were acquired, we only had 20some 19:57 engineers, so under 30 people serving 20:00 450 million monthly active users. So, we 20:05 didn't have code reviews. The only time 20:07 I got my code reviewed was the first 20:10 time I made a commit. Brian asked to 20:13 take a look at it before I committed it. 20:15 And he asked me a bunch of questions 20:17 which I had to think through a lot like 20:19 a kind of like a coding interview, but 20:22 that that was it. After the first time, 20:24 we didn't really have a formal code 20:25 review. But I mean, people read the git 20:28 commits because there's only 30 20:30 engineers. You can read other people's 20:32 code and they would discuss it on the 20:33 WhatsApp groups. 20:34 >> So, everyone was trusted, all engineers 20:37 that they just pushed their code to they 20:39 merged it into production, pushed it to 20:41 production without a manager review. And 20:44 it was trusted that, you know, they 20:45 would ask if they were unsure or 20:47 something like that. 20:48 >> Exactly. 20:49 >> Okay. And it worked. 20:51 >> It worked. What about the release 20:52 process? Like if if if you tell me 450 20:55 million people, the first thing I'm 20:56 going to say is like, okay, did you do 20:58 canarying? Did you do feature flagging? 21:00 Did you do experiments? Did you do you 21:02 know what kind of safety nets did you 21:04 have? Right. 21:05 >> We didn't do much of that, but we were 21:07 really big on dog fooding. So, every 21:09 time we were about to do a release, we 21:12 would all internally use it ourselves. 21:16 Yan, I think he might still say it on 21:18 his LinkedIn. If you look up Yan, he 21:22 said just quality engineer. 21:23 >> His title when he messaged me cuz I 21:25 didn't know who he was see it said chief 21:27 QA officer. 21:28 >> QA officer. 21:30 >> And I didn't know what that meant. I 21:32 thought it was some sort of weird joke 21:34 uh from the outside. So now it makes 21:37 sense. So he he he was going around. He 21:39 was making sure that it it worked. He 21:41 would try to break things as much as he 21:44 can and then if he finds a bug he will 21:47 like really try to break it and then 21:49 he'll come to it and say hey like I 21:51 found this bug 21:52 >> and you also said that Jan said no a 21:55 lot. He did say no almost as I recall 21:59 99% of the time he would say no which I 22:02 thought as a again as a young engineer I 22:05 was very confused because when you look 22:06 at all these other apps there were like 22:08 dozen different messaging apps at the 22:10 time like WeChat is notorious for having 22:13 everything right they have so many 22:14 features and I was so confused like why 22:17 don't we build all these features these 22:19 are the newest coolest things that we 22:21 should have because at the time when I 22:22 joined we didn't have groups We launched 22:26 groups shortly after I joined. We didn't 22:29 have voice calls, video calls. We didn't 22:31 have any of the no stories. You know, 22:33 all the cool features were missing in my 22:36 mind, but that was by design because we 22:38 really wanted to prioritize again the 22:40 quality of a grandma in a remote town 22:43 being able to use our app at any given 22:46 time. 22:46 >> WhatsApp how features for years until 22:49 they were absolutely sure about quality. 22:51 They worked on video calling long before 22:53 they shipped it. This leads us nicely to 22:55 our presenting sponsor, Stats Sig. 22:58 Today, you don't have to choose between 22:59 speed and confidence. Statig lets you 23:02 ship features behind flags, experiment 23:04 with real users, and only roll out 23:06 broadly when data shows that you're 23:08 ready. Here's what it looks like in 23:10 practice. You ship a change behind a 23:12 feature gate and you roll out gradually, 23:13 say to 1% or 10% of users at first. You 23:16 watch what happens. Not just did it 23:19 crash, but what did it do to the metrics 23:20 you care about? conversion, retention, 23:23 error rates, latency. If something looks 23:25 off, you turn it off quickly. If it's 23:27 trending the right way, you keep rolling 23:29 it forward. And the key is how the 23:31 measurement is part of the workflow. 23:32 You're not switching between three tools 23:34 and trying to match up segments and 23:36 dashboards after the fact. Feature 23:38 flags, experiments, and analytics are in 23:40 one place using the same underlying user 23:42 assignments and data. This is why teams 23:44 at companies like Notion, Brex, and 23:45 Atlassian use stats. Static has a 23:48 generous free tier to get started and 23:50 proicing for team starts at $150 per 23:52 month. To learn more and get a 30-day 23:55 enterprise trial, go to 23:56 static.com/pragmatic. 23:58 With this, let's get back to how Gan and 24:00 the WhatsApp team ship quality code with 24:02 close to zero formal processes. So, it 24:04 it sounds like WhatsApp had very very 24:06 little process. This was very very 24:09 interesting because when I worked at 24:10 Skype at the same time as you joined 24:13 WhatsApp and and I also joined in 2013 I 24:15 joined Skype and you joined WhatsApp in 24:17 2012. Skype was very proud that they 24:20 sent everyone to scrum training. I was a 24:22 scrum master other people scrum master. 24:24 So here we were with all the scrum, all 24:26 the consultants, all the everything and 24:29 WhatsApp out competed us with like a lot 24:31 smaller team and and no no scrum, no 24:34 TDD, no agile, 24:36 >> Skype, 24:37 >> 10,000 engineers. 24:38 >> Wow, that's a lot of people. 24:40 >> Yeah. 24:41 >> I mean, when you have a thousand people, 24:42 you kind of need these. 24:44 >> Yeah. Yeah. And and in all fairness, 24:46 like for example, one thing that this 24:47 whole scrum thing solved for a little 24:49 bit is we had more more than 100 teams 24:51 and everyone was working on different 24:52 things and because of all this 24:54 organization, we we had a prioritized 24:56 list of which teams are the most 24:58 important and those got all the support. 25:00 So I I guess one lesson might be that 25:01 when you're just big, it's just so much 25:03 harder to move fast and a small team can 25:06 out compete you. 25:07 >> Yeah. It just takes a long time even 25:09 just to communicate with everyone. Being 25:11 inside of WhatsApp, how did it feel to 25:13 see this massive growth? Not in your 25:15 team size, but but in the product usage, 25:18 the you know the people, the media, the 25:20 feedback. 25:21 >> We didn't have much media like nobody 25:24 knew about WhatsApp. 25:25 >> One interesting thing you told me about 25:27 the office is you had countdown 25:28 displays. Can you tell me about them? 25:30 What were these? What did it display? 25:31 Yeah. So, you you asked me a lot about 25:33 metrics and I think the really the only 25:35 metrics we track like we didn't really 25:37 pay too much attention to media or 25:40 Skype's usage numbers or other messaging 25:43 apps usage numbers, but the one metric 25:45 we counted down was number of days like 25:48 X number of days since the last outage. 25:51 >> Wow. No pressure. 25:55 >> Well, the numbers started to go up over 25:57 time. Maybe that helped to have it 26:00 visibly there. And when an outage 26:01 happened, do you remember what happened 26:03 after? Because these days in the tech 26:06 industry, it's all about blameless 26:07 pulsemortems. If an outage happens, you 26:10 know, we first mitigate it, then we get 26:11 together, then we write a document where 26:13 we try really hard to not say who push 26:16 caused this, but we come up with why the 26:18 system is like this and so on. How did 26:20 you go about like dealing with outages 26:22 and also following up and ensuring that 26:25 they don't they won't happen again? So 26:26 they I know they did these discussions 26:29 in the server group chats, but I wasn't 26:31 in the server group chats, so I can't 26:34 really say for sure. I mean, for sure, 26:36 we did not have documentations. 26:38 >> It sounds like a lot of things were 26:39 pretty simple. You talk with people, if 26:42 you have a problem, you try to fix it. 26:44 Don't overdo things for no reason. And 26:47 it seems to just work. and and then have 26:49 the key thing like if I guess if you put 26:51 out days since since outage people will 26:53 know like okay I should do what I can to 26:55 not have an outage 26:56 >> and everybody knew exactly who was 26:58 working on what so we didn't have to 27:01 blame anyone everyone just knew 27:03 >> WhatsApp was a massive massive massive 27:05 success what do you think made it so 27:07 successful in the early years and in the 27:10 es especially for for for the product 27:11 itself you know you've seen cacao you 27:13 were you were aware of some of the 27:15 competing messaging app what did 27:16 WhatsApp do that others did 27:18 There is a little bit of the networking 27:21 effect if you it's like the thing about 27:23 messaging app is that if you use it you 27:26 need your friends to use it and if your 27:27 friends use it you need to use it and 27:29 WhatsApp was the first to be on the 27:33 market that certainly helped but there 27:35 was a lot of competition but again I 27:38 think um a lot of other apps and 27:40 messaging apps were chasing features 27:43 thinking about adding the the shiniest 27:45 newest features whereas WhatsApp was 27:48 very intentional. They actually worked 27:49 on video calling for a very long time. 27:53 We were probably working on it by the 27:54 time you joined Skype when your founder 27:56 said we have video. Um we were working 27:59 on it but we just didn't launch it until 28:00 much later when we were actually like 28:02 100% sure about the quality of the the 28:06 feature. So we often held on to features 28:09 until we felt really sure before 28:12 launching them. 28:14 >> Interesting. because that is a little 28:15 bit of a different than the conventional 28:16 advice which is if you're a startup 28:18 launch early, get feedback, improve it, 28:21 and iterate it. It sounds like you did 28:22 the opposite. It's it's like polish it 28:25 and then do when you have full 28:26 conviction. 28:26 >> Yeah, we did use it internally. 28:28 Internally, we used the voice and the 28:30 video calling features with our 28:32 families. So, we had like a list. Okay, 28:35 like I have family members. These are 28:36 all my my parents and my brother and 28:38 sister's numbers. Let's enable it for 28:41 this beta group. and we used it for a 28:43 very long time before we launched it 28:45 with the public. 28:46 >> Two years into working at WhatsApp in 28:48 2014, Facebook announces their biggest 28:50 ever acquisition, WhatsApp, for $19 28:52 billion. What do you remember of this 28:55 time? How unexpected was it? And you 28:57 know what what what kind of feelings 28:58 what kind of emotions went through you 29:00 and the team around you? 29:01 >> I actually journaled soon after the 29:04 acquisition. So I looked up my journal 29:06 around this time 2014. So it's been over 29:08 10 years. But I looked at my journal and 29:11 I remember I was coding. I had this 29:14 Spotify playlist with noise cancelling 29:17 headphones. I had this playlist called 29:19 Let Me Think. This the one I I listen to 29:22 when I want to focus. And again like we 29:25 were in a pretty small office where I 29:27 can see everything. I was sitting in 29:30 pretty central location. So I I could 29:32 see people bustling and hustling which 29:35 was a little bit weird, but I tried to 29:36 tune it out so I can code. 29:39 But then from the side I saw um Nirage 29:43 who was the head of business at the 29:44 time. He was just like waving his arms. 29:47 He's he was a pretty tall guy so I could 29:49 see it. He's like like stop whatever 29:51 you're working on right now. Come into 29:54 the um we had one meeting room. 29:59 >> Come into the meeting room. And I was 30:01 like what is happening? Like we never 30:03 have meetings. Like we we 30:06 >> So you didn't have meetings? I mean, we 30:08 we have scheduled meetings every now and 30:10 then, but we rarely have like we we have 30:12 never had unscheduled meetings and we 30:15 rarely have meetings at all. So, I was 30:18 confused and I I dropped whatever I was 30:20 working on and I went into the 30:22 conference room and then they asked 30:24 like, "Turn off your phones." 30:26 >> WhatsApp, turn off your phone. That 30:27 that's kind of weird, right? 30:29 >> And I thought, "Oh my gosh, what's 30:30 happening?" Like, "Did we go out of 30:32 business?" 30:35 That was one thought. I thought are we 30:38 getting another raise of fun like round 30:40 of funding like a new investor coming on 30:43 board. It can't be that we sold the 30:45 company because Yan used to say he will 30:47 never sell the company. He used to 30:50 actually say uh selling your company is 30:52 like selling your baby. 30:54 And I remember we were waiting for quite 30:56 a while because there was one person 30:58 missing. 30:59 >> Oh. 30:59 >> And it turns out she was getting her 31:01 eyebrows done 31:03 >> with her phone tucked away. Yeah, she 31:06 came she came uh after the announcement, 31:08 but the news was about to hit the public 31:11 and they wanted to tell us before the 31:13 news hit and I I noticed that Yan and 31:17 Brian were making this face and I 31:20 couldn't tell what it was and then they 31:23 made the announcement WhatsApp has been 31:25 acquired by Facebook for $19 billion and 31:28 I I realized oh that was them trying to 31:32 hide their excitement. 31:34 That was a face. 31:35 >> Kind of smiley but not smiley. 31:38 >> And that was a really exciting moment. 31:40 And I I kind of zoned out for a little 31:42 bit because I was trying to remember, 31:44 hey, like how many shares did I get? 31:46 Like again, it was my first startup 31:49 ever. I didn't even negotiate my equity. 31:51 And honestly, I couldn't remember how 31:53 much equity I had. And I was trying to 31:56 think how much is a billion dollars? 31:57 That seems like a lot of money. And how 32:00 much is like 1% of 19 billion? I 32:02 couldn't do the math. And I I remember 32:04 sitting there thinking like trying to do 32:06 the math and then I thought, you know, 32:09 no matter how the math works, I think 32:11 one thing is clear. I'm gonna be rich. 32:15 >> And then Zuckerberg walked in. 32:17 >> Zuckerberg walked in to the meeting. 32:20 >> Yeah. 32:20 >> Wow. And then you had like a Q&A or 32:23 something. 32:23 >> We did. We did. Yeah. 32:24 >> What What kind of questions can you ask 32:26 at at this point or what kind of 32:28 questions did people ask? 32:29 >> There was a mix of excitement and 32:31 nervousness, right? um are we going to 32:33 have to change everything? Like because 32:36 I think a lot of the engineers were more 32:39 uh experienced and they talked about how 32:42 when Yahoo acquired companies, they 32:44 changed 100% and lost the what is it the 32:49 essence of the business. So there there 32:52 were a lot of questions around that and 32:54 Mark is actually very charismatic in 32:56 person and I thought he had great 32:58 answers at the time. uh he made sure 33:01 everyone feels assured that uh nothing's 33:04 going to change and he will try to 33:06 maintain it as much as possible. At 33:09 least that was the messaging at the 33:10 time. 33:11 >> Clearly this this was an amazing exit 33:15 and to this date it's not really been 33:17 repeated. May maybe a few companies 33:19 might have come close but definitely not 33:20 with with much such a small team. How 33:22 did you and and and your colleagues deal 33:25 with the fact that wow you've just got 33:27 an amazing financial exit but I guess 33:29 the company kind of continues inside of 33:32 Meta like it seems seems like you know 33:35 two things at the same time like okay I 33:37 have this like amazing financial exit 33:39 but there's also work how do I balance 33:42 how did you balance how do you decide 33:43 what next 33:44 >> that's twofold 33:47 so the the finance side in terms of that 33:50 aspect we actually got a lot of support 33:52 our uh business person organized many 33:56 meetings with like the accountants or 34:00 even a financial advisor. We invited a a 34:03 professor who was the founder of 34:05 Wealthfront and he gave us an hour of uh 34:08 finance advice and he recommended books. 34:11 Um I read the random walk down Wall 34:14 Street which is a great book. I 34:16 recommend people read it if you're 34:18 interested in financial management. And 34:20 I read several other books to really 34:22 educate myself to be able to manage this 34:27 new wealth that I I came across as a 34:30 young 29year-old. 34:32 >> Yeah. What changed to the day-to-day 34:33 once you officially became part of 34:35 Facebook? Did you have to move offices? 34:37 Did you know did you get a new title 34:40 added to like the the meta or chart? 34:42 That kind of stuff. 34:43 >> The changes were very slow in the 34:46 beginning. We didn't even move into the 34:48 meta or at the time it was called 34:50 Facebook headquarters uh Menlo Park 34:53 until at least a couple years after the 34:55 acquisition. So in the beginning 34:58 everything was same as usual. We still 35:00 had our old office. Well, we did 35:03 actually move to a little bit nicer 35:04 office, a slightly bigger office, but 35:08 other than that it was business as 35:10 usual. was Yan and Brian and we were 35:14 hiring but not you know at our similar 35:17 like so steady pace. 35:20 Um and I think not until when we 35:24 actually moved into the Facebook office, 35:27 we started seeing a little bit more um 35:30 cultural influence and merging like we 35:33 started using their like HR services or 35:37 recruiting services and things like 35:38 that. But it was a very gradual change 35:40 over time. 35:42 >> And then when WhatsApp became part of 35:44 Facebook, as I understand it, it it 35:46 still is even to this date its own 35:47 organization. like inside of Facebook, I 35:50 understand there's organizations like 35:51 Messenger or like there's the Facebook 35:54 uh group etc. So like did WhatsApp 35:55 remain its own kind of organization a 35:57 little bit shielded from the rest of 35:59 Facebook? 35:59 >> We had our own area. 36:01 >> Yeah. Or 36:02 >> WhatsApp and in the beginning we even 36:05 had like our own chairs and our own 36:08 whatever like walls and decorations that 36:10 we were using we brought them all over. 36:13 But over time, you know, there was more 36:15 and more mixing. 36:16 >> After the acquisition, 36:19 how did you started to hire more people? 36:21 How did the projects change? Did things 36:23 become more ambitious? Did you start to 36:25 add more features? Cuz clearly like you 36:26 were about 30 of you and then few in a 36:29 few years there was hundreds of people 36:30 working on WhatsApp. These days it must 36:31 be thousands of people like with those 36:34 people like what new work came up 36:36 because again originally WhatsApp was so 36:38 minimalist, right? And kind of so 36:39 scrappy. 36:40 >> I guess we were choosing to be small. 36:42 Not that there was not enough work for 36:45 us to do, right? So, one of the reasons 36:47 why we also tried to remain small was 36:49 actually Brian and Yan did not want to 36:51 raise too much money and it actually 36:54 cost a lot of money to serve so many 36:56 users. You have to pay for the servers. 36:59 You have to pay for the SMS registration 37:01 codes. Every year, Yan and Brian would 37:04 do uh all hands meetings. So, we did 37:07 have meetings 37:08 >> once a year. Uh and Brian was very 37:12 transparent. He will walk through our 37:15 earnings and expenses. 37:18 >> Interesting. Yeah, I had a lot of 37:20 information around this. So the three 37:22 main buckets of our spending was server 37:25 cost was about a third and then about a 37:28 third on salaries for the engineers 37:30 mostly and then a third uh the rest was 37:34 for the SMS fee. the when you try to 37:36 register you get that code and we have 37:38 to pay that 10 cents or whatever how 37:40 much it costs to send international 37:43 messaging 37:44 those numbers I mean they add up when 37:46 you have millions of people using your 37:49 app so they actually didn't want to grow 37:51 too fast because it gets very expensive 37:55 was free for the first year and then 37:57 after that WhatsApp was charging $1 for 38:00 every year but they were only using it 38:04 in certain countries trees really to 38:06 suppress growth because they don't want 38:08 to grow too fast. 38:09 >> Fascinating. Cuz I I remember in in in 38:12 Europe and in the US, there was this $1 38:14 cost which I think people were like, 38:16 "Yeah, well, whatever." I don't think we 38:18 realize that that this was a growth 38:20 discretion tactic. Fascinating. And then 38:22 when Facebook acquired, I guess they got 38:23 rid of it. 38:24 >> Yeah. Facebook said, "We don't need the 38:26 dollar. We can grow as much as we can 38:29 because they had the funding for it." 38:31 >> And then growth just did it. Did it 38:32 speed up? Do you remember? 38:33 >> It did. Yeah. incredible detail use 38:36 using payment to slow down growth. 38:38 >> The lesser known detail about the $1 is 38:41 that uh that $1 was enough to pay for 38:45 all of these the server cost, the 38:47 salaries and the SMS code 38:49 >> per year. So you were roughly break 38:51 even. 38:52 >> Break even. We did have funding from 38:54 Sequoia, but we never touched that 38:56 money. 38:57 >> Incredible. Yeah, Brian explained it as 39:00 how his dad was a business owner and 39:03 they would wake up in the middle of the 39:05 night worried what if I cannot pay the 39:07 the salaries for the employees tomorrow 39:10 and he he explained that he took the 39:13 funding from Sequoa as like a backup 39:16 >> and I think it was $8 million of funding 39:18 if I recall if I looked I looked at that 39:20 backup. 39:21 >> Yeah. So we never touched that money. 39:23 The $1 paid for everything 39:25 >> and it slowed down growth enough to be 39:26 manageable. Yeah. 39:27 >> When you joined Facebook, what what 39:30 title did you get? And how did your 39:32 career change? 39:33 >> So, the thing about Facebook is that 39:35 everyone's actually software engineer. 39:38 I'm pretty sure they still don't have 39:40 titles. 39:40 >> They don't have titles, but they have 39:41 levels. What What level did you come in 39:43 at? 39:44 >> So, being one of the five youngest 39:46 people, I got I got leveled as a junior 39:49 engineer. 39:50 >> No, you did not. L3 or L4? 39:53 >> L3. L3. Yeah. No, 39:56 >> I had to like climb climb all over 39:58 again. 39:58 >> Oh my gosh, that must have been a bit 40:00 awkward. 40:01 >> I was not too happy about it. But what's 40:03 the alternative? Do I want to give up 40:05 besting the rest of the shares? 40:07 And eventually I got promoted. 40:09 >> But it was with within WhatsApp. So you 40:11 got promoted pretty quickly. How many 40:12 times did you get promoted there? 40:14 >> A few times. I mean I eventually became 40:16 an engineering manager. 40:18 >> And then as you became an engineering 40:19 manager, uh at some point you decided to 40:22 help and start a new office in London. 40:25 How did that decision come and how did 40:27 you go about it? 40:29 >> That was actually uh an ask from 40:33 Facebook headquarters. So they said, 40:35 "Hey, like we're actually running out of 40:37 space in Menlo Park and also WhatsApp is 40:40 so big in Europe, so why not have a 40:42 presence there? It'll be much easier to 40:44 hire engineers because everybody 40:46 actually uses WhatsApp. So let's let's 40:49 start a new office there." And we didn't 40:51 have that many engineering managers, 40:53 right? I was very lucky because I got 40:55 asked to go along with couple other 40:58 engineering managers and all three of us 41:00 actually became managers around the same 41:02 time. We actually even trained together. 41:04 We were relatively new managers when we 41:06 got asked to go there. But I think we 41:08 were the only ones who could go because 41:10 you know people have children and they 41:12 have think about school and they they 41:14 couldn't go. I remember one the director 41:16 that I was working with he couldn't go 41:18 because his wife says she doesn't want 41:20 to move with the children. It it makes 41:22 perfect sense. You arrived in London, 41:24 you landed with these two or three other 41:26 engineering managers. How did you start 41:28 to grow the office from a practical 41:31 perspective? What can I imagine like you 41:33 know like how did you start hiring or 41:35 leasing space or what are the other 41:37 things that you had to do that you know 41:39 like were maybe a little bit unexpected 41:41 for you? 41:41 >> A lot of the logistical part was taken 41:44 care of for us because Facebook already 41:46 had an office there. So we kind of moved 41:48 in. We got our own section and it it 41:51 wasn't big because at the time again we 41:53 had a lot of contractors in Europe. So 41:55 we had one contractor already in 41:58 England. So we turned we uh converted 42:01 them full-time and then we had one in 42:03 Scotland. We also converted him 42:05 full-time so he would commute from 42:07 Scotland every now and then. So we had 42:09 two engineers plus three managers and we 42:12 started hiring there. I think the hiring 42:16 part was something that took longer to 42:19 set up. We worked very closely with the 42:21 Facebook hiring team, which was really 42:23 great that we already had people who are 42:25 familiar with the the local um 42:28 recruiting logistics there. So, one 42:31 thing we focused on a lot was really 42:33 letting engineers know, hey, WhatsApp is 42:35 hiring in Europe now. Come apply. 42:38 Because we were hiring from all over 42:39 Europe and also a lot from India. Do you 42:42 feel it was easier to hire for WhatsApp 42:44 in Europe just because people knew about 42:45 it? Do you get more excitement, more 42:47 applicants? 42:48 >> 100%. 42:50 You wouldn't believe like I used to do a 42:52 lot of university recruiting and when I 42:54 used to go to Stanford maybe 2013 42:59 like anytime before the acquisition I 43:01 would say hey like people will come up 43:03 to the booth and I would say hey do you 43:05 want to give me your resume and they 43:06 would be like tell me about your company 43:08 first. 43:11 because they they have never heard of 43:12 WhatsApp. What is this company? I'm not 43:14 even going to give you my resume. 43:15 >> My resume. I have only 20 of these. 43:17 >> Exactly. 43:19 >> Uh versus in Europe, people were 43:21 actually excited to talk to us. 43:23 >> What were the good and bad things of 43:25 working in what basically is a remote 43:28 office like yes, London was a big 43:29 office, but HH HQ was in California, 43:32 Menlo Park. That's 8 hours of time zone 43:35 difference. A lot less overlap. There's 43:36 probably some good things about this and 43:38 some downsides. It helped because the 43:42 three of us were from Menel Park and we 43:45 actually had great relationships with 43:47 other teams and other engineers and 43:49 other managers and we also traveled back 43:52 to Menel Park every quarter and then we 43:55 had the leadership from Menlo Park also 43:57 travel to London almost every quarter. 44:00 So there was a lot of back and forth um 44:03 to really strengthen the relationship in 44:05 the beginning. your your growth went to 44:07 like being I guess the one of the most 44:10 junior people in WhatsApp which is crazy 44:12 to say because you were experienced as 44:14 well but then you were also L3 and 44:16 Facebook which I still cannot believe 44:18 but you you you went and became a 44:20 manager. What pushed you to actually say 44:22 I actually want to try to manage people. 44:25 >> I actually never asked for it myself. 44:27 Someone on my team begged my manager, 44:31 hey can I please report to Jean? And 44:33 that's how I became a manager. Wow. 44:36 >> Okay. What do you think this this person 44:38 saw in you that they wanted to report 44:39 you when you were not a manager? 44:41 >> I was the tech lead. So I was already 44:43 managing the project. So it was sort of 44:45 a natural transition for me. 44:47 >> And when you became a manager, what 44:48 parts of the job came naturally to you 44:50 and what parts were hard that you had to 44:52 learn or get mentorship for? 44:54 >> You know, I started reading books. I 44:56 love reading books. Whenever there's a 44:59 new challenge, I like to read, learn, 45:01 and research. There actually at the time 45:04 weren't a lot of courses on how to 45:06 become a manager and not a lot of books. 45:09 Like I still don't think there are too 45:11 many books about how to become a 45:12 manager. 45:13 >> There's a little bit more now. There 45:14 there's like three or four good ones, 45:15 but but they all came out after like 45:17 2015 or 2016. Yeah, the the resources 45:20 were pretty limited, but I I did what I 45:24 can to read as much as I can about 45:27 leadership and I think I read read 45:29 actually a lot about communication and 45:31 psychology. There's several books like I 45:34 love the book surrounded by idiots. Have 45:37 you read that one? It talks about the 45:39 the disk personality, the different 45:42 types of personalities and I try to 45:44 really understand like what motivates 45:46 people, how do you communicate with 45:48 people in a in a way that makes sense to 45:51 the other person and also I reflected 45:54 personally like what were some good 45:56 managers and bad manager in my 45:58 experience because you hear the saying 46:01 that people don't leave companies they 46:02 lead managers right your manager can 46:05 really break or make your career and 46:07 they can make your life miserable if 46:09 you're, you know, matched with someone 46:11 you don't vibe with. 46:13 >> What are the traits that you found as 46:15 you recalled? What were things you said 46:17 like, I think this makes a good manager. 46:18 I want to do more of that and I think 46:20 these were terrible managers or bad 46:22 managers and I want to avoid doing that. 46:24 Do you remember some things that stuck 46:25 out? 46:26 >> Yeah, I tried to really understand each 46:29 individual person. So, for example, like 46:31 one person that I had on my team really 46:34 loves going deep into problems or 46:38 debugging and finding out how to improve 46:40 things, right? Whereas another person 46:43 really loves building new features and 46:45 you cannot ask this person who loves to 46:48 build new features to go debug 10 bugs 46:51 and that person will go nuts, right? And 46:53 then like one person who was really good 46:55 at uh building new features was not so 46:58 great at mentoring new colleagues. So I 47:00 try to really look for their strengths 47:03 and of course you also want to set them 47:05 up for challenges so they can learn as 47:08 well. But you want to balance them out. 47:10 So I I try to really understand by 47:12 asking them a lot of questions to 47:14 understand like how do they want to be 47:16 challenged? When do they feel excited 47:18 about their work or what are the things 47:20 that they're really good at? what are 47:22 the things they want to improve on? So, 47:24 I spent a lot of time really talking to 47:26 them. 47:26 >> As a manager, you were part of 47:28 calibration meetings, right? Now that 47:31 you're not at not at WhatsApp, not at 47:33 Meta, can we talk honestly about what 47:36 are those meetings like? Uh, you know, 47:39 what are maybe the the good things about 47:41 them? How how can you prepare and what's 47:43 the kind of reality? Cuz I feel outside 47:45 of a small group of managers who are in 47:47 there, it's not many people know like 47:49 how how these things go. So people 47:52 number one biggest mistake people make 47:54 is they think your manager is the one 47:57 giving you a promotion or a salary boost 48:01 like as a manager middle manager right 48:03 like I have no authority to give you a 48:06 promotion 48:06 >> you have no budget typically directors 48:09 have a discretionary budget sometimes to 48:11 be able to give a reward but not even a 48:13 promotions they even even they cannot 48:15 give right 48:15 >> right and um the bonuses are tied to 48:19 your performance review right So at Meta 48:21 for every level there's exact 48:23 percentages lined up by the comp team. 48:26 Like I have no control over it. The only 48:28 control I have is I think of myself as 48:31 the lawyer representing my clients. 48:34 >> Wow. Yeah. 48:35 >> I'm making a case for them. 48:37 >> Yeah. 48:37 >> Why they deserve to get a certain 48:40 performance review rating or a 48:42 promotion. And obviously like I want my 48:45 clients to do well. I want my team to 48:47 get you know the recognition that they 48:50 deserve because I know they worked hard 48:53 but it's not up to me. All the other 48:55 managers also have to agree that is the 48:58 the nature of performance reviews 49:01 >> and being specific on a performance 49:03 review like who were the people that you 49:06 saw the engineers who got these high 49:08 performance reviews from this committee? 49:10 What kind of tactics did you see? Were 49:13 there things where like well some 49:14 managers kind of like you know politics 49:16 were they kind of like they're calling 49:17 in favors for each other and pushing 49:19 someone up or or was it mostly 49:21 meritocracy meaning uh this engineer was 49:23 actually doing great work that a lot of 49:25 managers saw and they just naturally 49:28 agreed that you know this person who's 49:29 on on Jean's team is actually they 49:33 should be above my great person and I 49:35 kind of agree with that cuz because 49:36 there's bucketing right let's be clear 49:38 is bucketing you're going to have 49:39 buckets and you you need to put like I 49:41 don't know x people in the top bucket, 49:43 middle bucket, bottom bucket and so on. 49:45 >> Yeah. When I was coaching engineers, uh 49:49 I learned that different companies have 49:51 different ways of self-promotion. 49:55 So like for example, I heard some 49:57 companies use emails like they send mass 50:00 emails every time they do a new release 50:01 or launch or like at WhatsApp we use 50:06 WhatsApp groups for everything but at 50:08 Facebook they used Facebook workplace 50:10 which is like Facebook groups where you 50:12 have a group for team or your org and 50:15 your like everything has a different 50:17 group and I noticed as I'm representing 50:21 my clients during performance reviews 50:24 The people who post the most often, who 50:26 have the most visibility, 50:29 usually get the easiest consensus 50:33 because it's just like all very natural. 50:35 Like if I have no clue what you worked 50:38 on and your manager tells me you're 50:40 great, maybe, but how would I know? I 50:42 don't I don't know anything about you. 50:44 So, it I'm less likely to be inclined to 50:47 agree with your manager. Maybe your 50:49 manager is right, but I don't know. 50:51 Whereas if you have been actively 50:54 posting and telling me indirectly or 50:56 directly what type of work you have done 50:59 and what type of impact that has made 51:01 and what are the lessons that you 51:03 learned and what type of people you work 51:05 with then I already know oh okay like 51:07 when your manager tells me you're ready 51:10 then I I saw saying 51:11 >> and then internal wolf this was actually 51:13 like it's it's more than just groups it 51:14 was like this Facebook feed where you 51:16 know like it's a bit like LinkedIn right 51:18 just to make it so so you see these 51:20 posts come across the And sometimes you 51:22 will hit hit like and what you're saying 51:23 is like if you've seen this post from 51:25 this engineer on some other team saying 51:27 oh we've launched this feature here's an 51:29 interesting thing we've learned that 51:30 we're using for Facebook and I hit like 51:32 uh I now remember it and then when 51:34 performance review comes like oh I 51:36 remember that person they wrote that 51:38 >> exactly and I might even have some 51:40 questions right maybe like if your 51:42 manager tells me I might be like but 51:44 what about this what about that but if 51:46 you make a post I can just ask you 51:47 directly through the comments right 51:49 there's a lot of engagement happening 51:51 ing in the comments. So, I might ask, 51:52 "Have you thought about this other 51:54 thing? Have you thought about this 51:55 thing?" And you might give me answers 51:56 and I think, "Oh, okay. Yeah, he's 51:58 thought about it. He's really good." 51:59 >> It's amusing because it sounds like 52:01 simplifying a little bit. But to be 52:02 successful at Facebook, you need to also 52:04 be good inside of the Facebook app and 52:07 and do interesting work and and not hide 52:09 it, actually make it visible. 52:11 >> Mhm. 52:12 >> That's interesting. Now, stepping up a 52:14 step back and you were a manager at 52:16 Facebook. You saw a lot of engineers 52:18 outside of the performance review and 52:19 people posting about it. What traits did 52:22 the the best engineers that you remember 52:25 share? Like what made them so good? 52:27 >> I I struggle with this question a little 52:29 bit because there's a difference between 52:32 like how do you measure skill? How do 52:35 you measure what a good engineer is? Is 52:38 a good engineer someone who can bang out 52:41 new features? Is a good engineer someone 52:43 who can design a complicated system? Is 52:46 a good engineer someone who can 52:48 communicate all of this and explain it 52:50 to nontechnical people? I struggle a 52:53 little bit with the definition of a good 52:55 engineer because I can have a definition 52:57 of a good engineer, but it may be 53:00 different for every culture. Different 53:02 company might have different 53:03 definitions. 53:04 >> A good one at at Facebook. What was a 53:06 definition? I remember that a lot of it 53:08 went down to just a very simple 53:10 characteristic impact. Right. 53:12 >> Definitely and I think the way like 53:14 there are many ways to measure impact 53:17 and definitely at Facebook their way of 53:19 measuring impact was through these 53:21 posts. If I know about your work and you 53:25 tell me you have impact and I agree 53:27 that's impact. So going back to when you 53:30 were in London office and and start to 53:31 grow. At what point did the London 53:33 office start to feel less of a startup, 53:36 a scrappy startup and more of a big 53:38 tech? 53:39 >> I remember a time after about a year and 53:42 a half or so I realized I don't know who 53:46 that person is or I don't know their 53:47 name. 53:49 That was a turning point. 53:50 >> Mhm. And at what point did you actually 53:53 start to think of leaving Facebook? I I 53:57 think I really enjoy the intimate 53:59 environment. So, I appreciate being able 54:01 to like at WhatsApp with 30 engineers, I 54:05 knew everyone's names. I knew where 54:06 everybody lived. I knew their spouses 54:08 and their children, their dogs names, 54:10 right? 54:12 I really like that type of intimate 54:14 environment. Um, we still hang out that 54:16 we have a pretty strong bond. And I feel 54:19 like when when I even when I don't even 54:21 know this person's name, I I just feel 54:25 less connected. 54:26 >> Yeah. So So was this the point where you 54:29 decided that maybe it's time for you to 54:30 leave and do something else? 54:32 >> Oh, so okay. I was um in London on a 54:35 contract. So I had a 2-year contract. 54:38 They said, "Hey, like go start this 54:40 office." And then once the contract 54:42 ended, I had the option to either stay 54:45 there to continue working in the London 54:47 office or I could come back to Men Park. 54:52 But then at that point, I had been 54:53 working there for 8 years. And honestly, 54:55 I think I was pretty burned out. I'm the 54:58 type of personality who likes to get 55:00 like A+ on everything I do every single 55:03 time. 55:04 >> Yeah. 55:05 >> So it was pretty tiring after 8 years. I 55:08 needed a break. 55:08 >> Yeah. And when you left WhatsApp, what 55:11 did you decide to do? What I say 55:14 WhatsApp but it was Facebook at that 55:16 point. 55:16 >> Yeah. Um I actually because I know my 55:19 personality I don't take breaks. 55:23 So I actually had a goal. This is simple 55:26 but I said I will do nothing for the 55:29 next 6 months. I'm going to challenge 55:31 myself to do nothing for 6 months. 55:34 >> Did you manage? 55:36 >> I did it. I did it. I did read a lot. I 55:39 exerc exerciseed. I went on long walks. 55:41 I did multiple meditation retreats. But 55:44 that that was my challenge to myself to 55:46 not work for six months. 55:48 >> So after 6 months of successfully doing 55:50 nothing, after setting yourself that 55:52 goal, what did you do to figure out what 55:54 next? 55:55 >> So initially I thought maybe I want to 55:58 go start a new company or join another 56:00 startup because I like working. I love 56:03 building things. So I decided, okay, I'm 56:06 going to start talking to other founders 56:08 or people who are hiring or people who 56:10 are looking to start a new company. So I 56:12 I actually talked to 100 founders. I 56:15 have a spreadsheet. 56:16 >> Wow. 56:17 >> To really see like is there any 56:19 interesting opportunities that I might 56:20 feel passionate about joining or 56:22 building. Then after talking to 100 56:25 startups, I realized I wasn't really 56:27 passionate about joining any of them. 56:29 And I thought like what what would I 56:31 feel more passionate about and what was 56:33 the thing that I liked the most about 56:35 working at WhatsApp for the past eight 56:37 years. And I realized I actually really 56:39 liked being a manager because I felt 56:41 like I was creating a culture of like 56:45 support so that other people can really 56:48 be learning and thriving and you know be 56:51 able to do things freely without people 56:54 breaking down your neck or there are 56:56 many things that make for a happy 56:58 career. But I found it really um 57:00 gratifying to be able to find that from 57:03 each person and really try to help them 57:05 out and create whatever that is. It 57:08 might be different for different people 57:09 and trying to unblock them so they can 57:11 really flourish. And I thought, well, if 57:13 that's what I really want to do, I don't 57:15 have to start a new company. I'll just 57:17 do that part. So, I started exploring 57:20 like mentoring people. Um I did a little 57:23 bit of coaching I don't do anymore. 57:25 making videos on YouTube, writing um all 57:28 of that to see how how would I find the 57:32 best way to support other people 57:34 >> and on on YouTube and on LinkedIn you 57:36 have been sharing a lot of your 57:37 learnings, your observations. What what 57:39 pushed you to to start sharing way more 57:42 than before like I I think you started 57:44 to do this publicly after you left 57:46 Facebook. 57:46 >> I was actually writing a blog about 57:48 this. 57:50 So, I actually just hit 100K subscribers 57:53 on YouTube like last week. Thank you. 57:56 Um, and I was reflecting 57:59 I almost gave up doing YouTube because I 58:03 was really not comfortable being seen in 58:06 public. And I I I've been thinking a lot 58:08 about this. Like my grandma's from North 58:10 Korea. She escaped during the war. And 58:14 in that culture, like you are you do not 58:17 speak publicly. 58:18 um you don't want to be seen because 58:21 it's dangerous and I think there's 58:24 generations of that still kind of 58:26 installed in me. The the fear of 58:28 speaking up is real. I felt really 58:31 uncomfortable. So I almost stopped doing 58:33 YouTube uh once one of my videos went 58:35 viral from early on and I felt really 58:38 uncomfortable. But luckily I was talking 58:40 to a mentor of mine and she said, "Hey, 58:43 it's okay to do something that you enjoy 58:45 doing. Just give it a shot." So then I I 58:48 stuck with it. I'm so glad I did. 58:50 >> Speaking of the thing that is happening 58:52 of course right now, AI, you you spoke 58:55 about this on on your YouTube channel as 58:57 well, but from your your vantage point, 59:00 how is AI changing how engineers work, 59:03 how managers work? 59:04 >> I do find it really interesting how with 59:07 AI, we're seeing smaller teams emerge. I 59:12 know that a lot of teams are saying, 59:14 well, we're small because of AI. But I 59:16 wonder if it's independent from AI. When 59:20 you're small, you're just more efficient 59:22 because WhatsApp did not use AI, 59:26 but we were efficient because we were 59:29 small. 59:30 >> And I almost feel that even today, I 59:34 can't cannot really point to too many 59:36 teams that are as small as WhatsApp and 59:39 have that kind of impact. Maybe and 59:42 might come to mind, but I think even 59:43 they're bigger. So I I wonder if if 59:45 there is a maybe just going back to 59:48 basics with all of us, maybe AI allows 59:51 to do the way most companies would have 59:54 wished they operated. 59:55 >> Yeah. And I think there's also a shift 59:57 in the mindset like I remember back in 59:59 the days people when you go to 60:02 networking events, people would brag 60:04 about oh like we've hired like a 60:06 thousand new engineers or we're growing 60:09 at X times bigger and that was like a 60:12 point of brag. And investors also 60:14 thought that was a good thing. Like you 60:15 need to grow, you need to hire more 60:17 engineer. That was a sign of healthy 60:19 engineering environment. Whereas 60:21 nowadays 60:23 investors actually think smaller is 60:25 better, right? Like they don't 60:27 necessarily push you to hire more 60:29 people. And I think as a byproduct of 60:32 hiring less people and staying lean, 60:34 they have found this new found 60:36 efficiency and they happen to equate it 60:38 with AI. Although AI I think it's clear 60:41 it makes engineers a lot more efficient 60:43 uh in well we think it makes them 60:46 efficient because it can generate a lot 60:47 of code you can work on more things 60:49 parallel is happening with agents how 60:51 are you seeing the role of software 60:53 engineers change and also the role of 60:54 engineering managers 60:56 >> yeah I mean I love AI tools I I use it 60:58 every day as a thought partner I I often 61:02 ask chacha hey like be my executive 61:05 coach or be a hardware trained futurist 61:08 and and help me find the next trends or 61:12 you know there there are various ways of 61:13 really using AI to its full potential. I 61:16 feel like engineering management is less 61:19 affected by AI because it it requires a 61:22 lot of like peopleto people like asking 61:25 questions and learning about your 61:26 engineers. AI can maybe help you with 61:29 that but I don't see AI replacing that 61:32 part. 61:33 But again because the teams are much 61:35 smaller if you were the type of 61:37 engineering manager who was doing a lot 61:39 of these like OKR and process and 61:42 writing documentation a lot of that part 61:44 is going to be gone and I'm kind of glad 61:47 it it will be gone because I I don't 61:49 think it's really necessary. 61:51 >> Yeah. For example, a lot of performance 61:52 management of you know gathering the 61:53 impact it can probably be done by asking 61:56 agents to gather all these things. I 61:58 remember as an engineer manager I used 61:59 to go through gathering all the work 62:01 that my engineers have done. So on the 62:04 calibration meeting I could fairly 62:05 represent them and then turns out that 62:07 the managers who showed up without doing 62:08 that I had an advantage but that was not 62:10 fair for the engineers by the way right 62:12 maybe AI will get rid of this advantage. 62:14 Yeah, AI will do a lot of the the grunt 62:17 work, more tedious work that maybe 62:19 engineering managers or even software 62:21 engineers had to do manually back in the 62:23 days like uh we had an engineer who was 62:26 just there to add comments. 62:29 I think that is something AI can do 62:31 really well. 62:31 >> If you had to give career advice to a 62:33 new grad who says I would like to build 62:35 a durable career in software engineering 62:38 in this kind of AI native world, what 62:40 would you suggest they focus on? I say 62:43 foundations, 62:45 you know, tools come and go, languages 62:48 come and go, but foundations don't go 62:50 anywhere. 62:51 >> We mentioned that at WhatsApp that's 62:53 WhatsApp was very small, very efficient. 62:55 What do you think today's AI or like AI 62:59 native startups could still learn from 63:01 WhatsApp that made WhatsApp successful 63:03 and it probably helped them as well. I 63:05 think of AI. So like we went through 63:08 several trends like when I first got my 63:11 first internship ever, it was video 63:12 sharing website and I've seen how there 63:14 were dozens of video sharing websites 63:17 and how the ecosystem changed over time 63:20 and then I saw with WhatsApp there were 63:22 dozens of other messaging app 63:24 competitors and how that kind of settled 63:27 down over time. I think we're living 63:29 through something similar. There are so 63:31 many new AI startups and new tools and 63:34 so easy to get distracted by all the 63:37 different options and it can feel quite 63:39 overwhelming. There are too many options 63:40 and you can feel the decision paralysis 63:44 but really again go back to the core 63:46 foundation. Think about like if you're a 63:48 builder think about what you're 63:49 building, why you're building. If you're 63:51 learning, think about why you're 63:52 learning, what you want to learn, and if 63:54 you have clear goals of where you want 63:56 to go, it will really ground you because 63:58 otherwise you're just going to be all 64:00 over the place and you might work really 64:01 hard and end up nowhere. 64:03 >> Do I understand correctly that you're 64:05 saying that WhatsApp was successful 64:06 because the goal was clear. Jan said no 64:10 to the distractions and all the ideas, 64:12 but it was very clear marching. Whereas 64:14 all the other competitors, even all the 64:16 messaging apps, they got distracted 64:18 building. Oh, let's do like oh this cool 64:20 video feature. Let's do stories. Let's 64:21 do all of these things. They saw 64:23 traction and they did a lot of these 64:24 things. But WhatsApp was very good at 64:27 doing the core thing well and then 64:29 slowly adding things that were a value 64:32 added. Is that a fair summary? 64:34 >> Yeah. And also I noticed this when I 64:36 started advising and coaching uh startup 64:39 founders as well. And also for any 64:41 engineers who want to join new startups, 64:43 this is great way to evaluate new 64:46 founders. like some founders if you're 64:47 the opposite of yan and saying no to 64:50 things um that I call it removing 64:53 distractions right you're prioritizing 64:55 ruthlessly if you're the opposite of 64:57 that imagine what type of startup you 64:59 end up you say yes to everything maybe 65:01 it might feel really nice as a 20some 65:04 year old if I were to go back in time 65:06 and I go to the founder with all my 65:08 great ideas and he says that's a great 65:10 idea gene let's build it but imagine 65:12 like he said that to every single idea 65:14 that I had the company will be all over 65:16 the place in terms of the long-term 65:19 growth. It's not a very ideal situation. 65:22 >> So looking back, what are some kind of 65:24 like preI or not as modern practices 65:28 that you did at WhatsApp that were 65:29 really good that today's very modern AI 65:32 native teams or whoever could benefit 65:34 from. 65:34 >> Yeah, of several things come to my mind. 65:37 I think one of it is by having lean 65:40 teams, you get several benefits. You get 65:43 to remove a lot of the distractions and 65:45 process and through that you get two 65:48 really incredible benefits which is 65:51 ownership and the the really like the 65:54 freedom to build things 65:57 right because Jan was always like Jan 66:00 and Brian were always very specific 66:02 about what we're building 66:06 but how we're building it was up to 66:09 debate right I mentioned earlier that 66:11 the only time we did a actual code 66:14 review was the first time I made my git 66:17 commit. Uh Brian reviewed my code and 66:19 asked me a bunch of questions. So Jan 66:20 and Brian were both like so technically 66:23 adept. They were really excellent at 66:25 doing this. They would ask we're trying 66:26 to achieve this like what is the problem 66:28 here or what is the best way to solve 66:31 this issue? What are like different ways 66:33 we can approach this? Tell me. So, so do 66:36 I understand correctly that of course 66:38 the small teams help with a lot of 66:39 things but then having the founders push 66:43 people they hire especially early on it 66:46 almost like push them to excellence 66:48 right is it fair to say that by Brian 66:50 doing that super detailed code review 66:52 with you the first time it it just upped 66:54 your game and later he didn't even have 66:56 to do anything right 66:58 >> yeah and there's like multiffold right 67:01 like one is to really challenge me to 67:03 think critically and then I took I I 67:06 learned a lot just from that 67:07 conversation. And then also like from 67:10 then on he never checked my code again. 67:12 So I know I am responsible right and I 67:16 do believe when you give 67:18 responsibilities to people people will 67:20 step up. I mean not everyone 67:23 >> but most people will 67:24 >> but I think this might be a bit 67:25 underrated. I I wonder if we've had a 67:27 little bit of two over babying of 67:29 engineers. I I remember for a long time 67:32 there was this talk in the you know in 67:34 the past 5 to 10 years in the as engine 67:36 managers like well I have a new grad 67:39 will take them months to on board. I 67:41 need to sign up a mentor for at least 6 67:43 months maybe even a year and where are 67:45 we over babying these very capable 67:47 adults you know they're adults right 67:49 even if even if they're 18 but they're 67:51 typically 20some because they came out 67:53 of college and they're hungry and 67:54 they're ambitious and maybe we don't 67:57 need to do as much of it always. 67:59 >> Yeah. I think as long as you hire smart 68:02 people, it's kind of like a mold, right? 68:04 If you make a mold too small, that's 68:06 that's only the limit of how far they 68:09 will grow. 68:09 >> Yeah. If the mold is too small, you have 68:12 to throw away a lot of things that could 68:14 have made excellent material. And 68:16 finally, you're a reader. What are some 68:19 books that you would recommend for 68:21 software engineers or people wanting to 68:22 grow professionally or in a personal 68:24 sense? 68:25 >> I love reading books. I did um so while 68:28 I I challenged myself to do nothing, I 68:30 actually read I actually took a year but 68:33 I did read a 100 books during that time. 68:35 That was my doing nothing. Anyways, um 68:38 it kind of depends of what your goals 68:39 are, but you gave me some specific 68:41 things like for your career. I think for 68:44 me what was really helpful was what 68:47 color is your parachute? That helped me 68:50 really understand my strengths and my 68:52 goals and priorities in my career and 68:54 life. I mentioned the book surrounded by 68:57 idiots. I know the title is kind of 68:58 funny, but it's an excellent book if you 69:00 want to learn more about how to really 69:02 communicate and work with different 69:04 people. If you want to understand 69:05 finance, I mentioned earlier the random 69:07 walk down Wall Street. It's a great book 69:09 for understanding how to manage your 69:11 money. Yeah, I I would recommend those 69:13 books to start with. 69:14 >> And any fiction books? 69:15 >> Hunger Games was one of my favorite 69:17 books. I I read the whole series. 69:18 >> I I read it as well and I I almost like 69:21 the I like the movies as well, but I 69:23 love the books. Yeah. Yeah. I I love the 69:26 story of like this woman overcoming her 69:28 challenges 69:29 >> and everyone else and winning in the end 69:31 several times. 69:32 >> Yes. 69:33 >> Well, Jean, thank you so much. 69:35 >> Yeah. Thank you. Thank you for having me 69:37 on the channel. 69:38 >> This was a great conversation. 69:40 >> I hope you enjoyed this rare 69:41 conversation with Jean. One thing that 69:43 stuck with me was Jean's point about why 69:45 WhatsApp had almost no process and why 69:47 it worked. Processes exist for audits, 69:49 for accountability, and for tracking who 69:51 did what. But when you have 30 people 69:54 and everyone can see what everyone else 69:55 is working on, you don't really need a 69:57 paper trail. You just walk over and 69:59 talk. This is a good reminder that most 70:01 processes exist to solve problems that 70:03 are created by scale and not by the work 70:05 itself. I also found the Skype contest 70:08 really surprising. A thousand engineers, 70:10 Scrum certifications, twoe sprints, and 70:12 a dedicated scrum master for every team. 70:14 I was one of them at Skype. and WhatsApp 70:17 with 30 people and zero for my 70:18 methodology was shipping faster and 70:20 growing faster on every metric that 70:22 mattered. This is a much needed reminder 70:24 that organizational discipline and 70:26 actual shipping speed are just not the 70:28 same thing and I was in the middle of 70:30 this at Skype and Gene was at in the 70:32 middle of it in WhatsApp. Finally, it 70:34 was interesting as a former manager to 70:36 hear how Gene described performance 70:37 reviews as a manager herself. She 70:39 described herself as a lawyer 70:41 representing her clients, as in she 70:43 doesn't control the promotion. She just 70:45 makes the case. And the engineers who 70:47 had the easiest time getting promoted 70:48 were not necessarily the best engineers. 70:51 They were the ones who made their work 70:52 visible. They posted about their 70:54 launches in the internal Facebook 70:56 workplace. They engaged in comments, 70:58 answered questions publicly, and the 70:59 managers in those calibration rooms are 71:01 making decisions about people that they 71:03 might have never worked with directly. 71:04 So visibility is not just vanity. It's 71:07 how the system inside larger companies 71:09 actually works. This is an uncomfortable 71:11 truth, but I think every engineer at a 71:13 big company needs to hear it. If you've 71:15 enjoyed this podcast, please do 71:16 subscribe on your favorite podcast 71:18 platform and on YouTube. A special thank 71:20 you if you also leave a rating on the 71:21 show. Thanks and see you in the next