Latest / The Pragmatic Engineer / “The Coding Machine” at Meta with Michael Novati

Transcript

“The Coding Machine” at Meta with Michael Novati · The Pragmatic Engineer

• 0:00
  SPEAKER_0: What does a coding machine mean?
• 0:01
  SPEAKER_0: It's very controversial because anyone can write a lot of code in theory.
• 0:05
  A junior engineer could sit down and just press the keyboard all day and write
• 0:07
  a lot of code.
• 0:08
  SPEAKER_0: The zoomed out thing is impact.
• 0:10
  SPEAKER_0: So, what is the code doing?
• 0:12
  SPEAKER_0: Writing a lot of code all day long, but is it moving projects forward?
• 0:16
  Is it helping launch a product that might need a team of 5 to 10 engineers that
• 0:20
  just one engineer can do?
• 0:22
  SPEAKER_0: Is it unblocking a lot of people by just pushing a lot of things forward?
• 0:25
  Is it refactoring a lot of code extremely quickly and fast so that new
• 0:30
  engineers don't even have to deal with old libraries?
• 0:32
  These are the type of examples that make a coding machine a coding machine.
• 0:35
  SPEAKER_0: I'm still a coding machine to this day at Formation.
• 0:37
  SPEAKER_0: I'm writing code all day long.
• 0:39
  SPEAKER_0: I think I have 20 commits today already.
• 0:41
  SPEAKER_1: Oh, wow.
• 0:41
  SPEAKER_1: And it's just a morning.
• 0:43
  Michael Novati spent eight years working at Meta from 2009 to 2017,
• 0:46
  where in some years he was the number one code committer across the whole company.
• 0:51
  He started as an intern and in just six years was promoted four times getting
• 0:55
  to the E-7 level,
• 0:56
  which is the equivalent of principal engineering level at other tech companies.
• 0:59
  Michael is now the co-founder of Formation,
• 1:01
  our remote software engineering fellowship program.
• 1:04
  In today's episode,
• 1:04
  we go into what the coding machine archetype is as Meta for staff plus
• 1:08
  engineers and how Michael became the first ever coding machine at Facebook.
• 1:12
  Meta's engineering culture and why the company built so many custom
• 1:15
  internal tools.
• 1:16
  How the hiring committee works at Meta and how they make hiring decisions?
• 1:20
  And a lot more advice for doing well on tech interviews and additional
• 1:23
  stories from Michael's time at Meta.
• 1:25
  If you enjoy the show,
• 1:26
  please subscribe to the podcast on any podcast platform and on YouTube.
• 1:31
  SPEAKER_1: Michael, welcome to the podcast.
• 1:33
  SPEAKER_0: Hi, I am happy to be here.
• 1:34
  SPEAKER_0: Thanks for inviting me.
• 1:36
  So,
• 1:38
  let's start with something really surprising and interesting about you.
• 1:42
  SPEAKER_1: At Meta, there's these things called archetypes.
• 1:45
  They're used for performance reviews and promotions as I've learned when I
• 1:48
  researched the company,
• 1:49
  things like fixer text lead and there's something called the coding machine,
• 1:53
  which apparently came about because of you.
• 1:58
  SPEAKER_1: Can you talk about that?
• 2:00
  SPEAKER_0: Yeah, sure.
• 2:01
  I can talk a little bit about the coding machine,
• 2:03
  so it's back in maybe 2015 or so.
• 2:08
  So in general about the archetypes,
• 2:11
  there's two purposes for them at Meta.
• 2:15
  SPEAKER_0: Overall, fairness is really important at Meta.
• 2:19
  So the archetypes are these examples of what an engineer who's at the
• 2:25
  principal senior staff level would look like,
• 2:29
  and the archetypes are modeled after existing engineers who are there.
• 2:33
  So they're a way for the company to pattern match upcoming engineers against
• 2:39
  these existing engineers and try to have a fair system for who's at this level
• 2:44
  and who's not.
• 2:45
  And the reason why there's not just one or two,
• 2:50
  and there's a handful of them,
• 2:51
  is that Meta has this...
• 2:53
  They don't use this analogy a lot internally,
• 2:55
  but I've always found this a natural analogy is the pro sports team analogy.
• 3:01
  When you have a professional sports team,
• 3:02
  you have a bunch of baseball players say,
• 3:05
  or basketball players,
• 3:06
  and each person has a different position that they play.
• 3:09
  SPEAKER_0: But all of those people at the pro level are pretty good.
• 3:12
  SPEAKER_0: They're all quite good at every position.
• 3:14
  SPEAKER_0: They can throw baseballs, throw basketballs, etc.
• 3:17
  SPEAKER_0: Steph Curry is the best three point shooter I think.
• 3:21
  I don't know that much about basketball,
• 3:22
  and in the analogy of the archetypes,
• 3:24
  there's similarly people have specialties.
• 3:27
  So everyone,
• 3:28
  once you're at that level,
• 3:29
  you can do all the basic things engineers to do or on a daily basis,
• 3:33
  but there becomes these different specialties where you're exceptional or
• 3:37
  one of the best in the company or even best in the industry at.
• 3:41
  So the coding machine is something that it was invented for me,
• 3:48
  and I won't go into all the history of the back and forth of how it happened,
• 3:52
  but generally speaking,
• 3:53
  there wasn't really an archetype for what I was doing already.
• 4:00
  So this was a case where they had to sit down,
• 4:03
  compare me to people who are already at these E-7 plus levels,
• 4:08
  zoom out and focusing on impact.
• 4:10
  That's like the word you'll hear a lot when you talk to Facebook people about
• 4:15
  performance reviews is impact.
• 4:17
  And they try to say,
• 4:18
  they notice that overall Michael is impacting the company very similarly to
• 4:23
  other E-7s,
• 4:24
  but there isn't an archetype that he fits into.
• 4:26
  SPEAKER_0: I'm not a specialist or a fixer is a really good example.
• 4:32
  I don't write one line of code to save $10 million,
• 4:34
  simple cut,
• 4:34
  dry,
• 4:36
  easy,
• 4:36
  [inaudible 00:04:37].
• 4:37
  SPEAKER_1: So see that's what a fixer is.
• 4:38
  SPEAKER_0: Yeah.
• 4:39
  I have a friend who's a fixer and he writes a few lines of code a year and they're
• 4:43
  extremely impactful because he's spending most of his time talking to
• 4:47
  people and deep diving into these really weird intersections of issues.
• 4:51
  SPEAKER_0: And that's pretty easy to quantify.
• 4:53
  SPEAKER_0: It's easy to understand the impact on the business.
• 4:58
  SPEAKER_0: So, some of those archetypes are a little easier.
• 5:00
  SPEAKER_0: Now with coding machine it's a little harder.
• 5:02
  It's very controversial because anyone can write a lot of code in theory,
• 5:07
  like a junior engineer could sit down and just press the keyboard all day and
• 5:11
  write a lot of code.
• 5:12
  SPEAKER_0: So, what does a coding machine mean to be at this level?
• 5:16
  SPEAKER_0: And the zoomed out thing is impact.
• 5:21
  SPEAKER_0: So, what is the code doing?
• 5:24
  SPEAKER_0: Writing a lot of code all day long, but is it moving projects forward?
• 5:29
  Is it helping launch a product that might need a team of 5 to 10 engineers that
• 5:34
  just one engineer can do?
• 5:37
  SPEAKER_0: Is it unblocking a lot of people by just pushing a lot of things forward?
• 5:42
  Is it refactoring a lot of code extremely quickly and fast so that new
• 5:47
  engineers don't even have to deal with old libraries?
• 5:51
  So these are the type of examples that make a coding machine a coding machine.
• 5:58
  Now,
• 5:59
  how they created this,
• 6:00
  there's maybe the story behind that is just a lot of back and forth.
• 6:03
  It's an executive who really sponsored me,
• 6:08
  his name's Tom Allison and I think that he's head of Facebook,
• 6:12
  "Theproduct"now-
• 6:15
  SPEAKER_1: Wow.
• 6:15
  SPEAKER_0: ... or the app they call it, I think.
• 6:17
  Yeah,
• 6:18
  he was really championing me and working with the other VPs at the time to
• 6:24
  connect the dots and make this happen.
• 6:26
  SPEAKER_0: Yeah.
• 6:27
  SPEAKER_0: And I'm still a coding machine to this day at Formation.
• 6:30
  SPEAKER_0: I'm writing code all day long.
• 6:31
  SPEAKER_0: I think I have 20 commits today already.
• 6:34
  SPEAKER_0: This is a busy day.
• 6:36
  SPEAKER_1: Oh wow.
• 6:36
  SPEAKER_1: And it's just the morning.
• 6:38
  SPEAKER_0: Yeah, just this morning.
• 6:42
  SPEAKER_0: Yeah.
• 6:42
  I would say,
• 6:43
  in terms of going back to that professional sports analogy,
• 6:48
  that's my thing that I'm the top 1% at.
• 6:51
  On the professional sports team,
• 6:52
  it's like we need to move this new prototype forward insanely quickly.
• 6:58
  Then I might be the go-to person,
• 6:59
  or we need to refactor a lot of code really fast,
• 7:02
  I'm the one person you would go to and the whole company to talk to about doing that.
• 7:06
  SPEAKER_0: And that's what justified me being at that level.
• 7:09
  SPEAKER_1: Trust isn't just earned, it's demanded.
• 7:12
  Whether you're a startup founder,
• 7:13
  navigating your first audit or seizing security professional skill in your
• 7:17
  governance risk and compliance program,
• 7:19
  proving your commitment to security has never been more critical or more complex.
• 7:23
  SPEAKER_1: That's where Vanta comes in.
• 7:25
  Vanta can help you start or scale your security program by connecting with
• 7:29
  auditors and experts to conduct your audit and set up your security program quickly.
• 7:33
  Plus,
• 7:34
  with automation and AI throughout the platform,
• 7:36
  Vanta gives your time back so you can focus on building your company.
• 7:40
  Businesses use Vanta to establish trust by automating compliance needs
• 7:43
  across over 35 frameworks like SOC2 and ISO 27001.
• 7:48
  With Vanta,
• 7:48
  they centralize security workflows,
• 7:50
  complete questionnaires up to five times faster and proactively manage
• 7:53
  vendor risk.
• 7:55
  Join over 9,000 global companies to manage risk and prove security in real time.
• 7:59
  For a limited time,
• 8:00
  my listeners get $1,000 off Vanta at vanta.com/pragmatic.
• 8:05
  SPEAKER_1: That is V-A-N-T-A.com/pragmatic for $1,000 off.
• 8:11
  SPEAKER_1: This episode is brought to you by WorkOS.
• 8:13
  If you're building a SaaS app,
• 8:14
  at some point your customers will start asking for enterprise features like
• 8:17
  SAML authentication,
• 8:18
  skin provisioning and fine-grained authorization.
• 8:21
  That's where WorkOS comes in,
• 8:23
  making it fast and painless,
• 8:24
  add enterprise features to your app.
• 8:26
  Their APIs are easy to understand and you can ship quickly and get back to
• 8:29
  building other features.
• 8:31
  WorkOS also provides a free user management solution called AuthKit for up
• 8:35
  to 1 million monthly active users.
• 8:37
  It's a drop-in replacement for Auth0,
• 8:39
  income standard with useful features like domain verification,
• 8:42
  rule-based access control,
• 8:43
  bot protection and MFA.
• 8:45
  SPEAKER_1: It's powered by Radix, which means zero compromises in design.
• 8:49
  You get limited customizations as well as modular templates designed for
• 8:52
  quick integrations.
• 8:54
  Today,
• 8:54
  hundreds of fast-growing startups are powered by WorkOS,
• 8:57
  including ones you probably know,
• 8:58
  like Cursor,
• 8:59
  Vercel,
• 9:00
  and Perplexity.
• 9:01
  SPEAKER_1: Check it out at workos.com to learn more.
• 9:04
  SPEAKER_1: That is workos.com.
• 9:07
  Do I understand correctly that A,
• 9:09
  this happened when you were at E-7 level,
• 9:11
  which is a very senior level,
• 9:12
  I think equivalent of senior staff or principal engineer at companies that
• 9:16
  do have levels.
• 9:19
  And then B,
• 9:21
  can you give an example of what type of work you were doing at the time?
• 9:24
  SPEAKER_1: Just give us an example of what your week-to-week looked like.
• 9:27
  SPEAKER_1: Because you mentioned refactoring or doing impactful stuff.
• 9:31
  SPEAKER_0: Yeah.
• 9:32
  So the conversations start when you're at the E-6 level,
• 9:36
  which is generally considered staff.
• 9:41
  Three is junior,
• 9:42
  four is mid-level,
• 9:44
  five is senior,
• 9:45
  six is staff,
• 9:46
  and then seven is senior staff.
• 9:48
  I mean,
• 9:48
  getting 3,
• 9:48
  4,
• 9:48
  5 is your more standard common traits between those levels.
• 9:55
  And at six,
• 9:56
  you start to think about what makes me a little different than others.
• 9:59
  You're participating at a higher level and then that's when you have the
• 10:02
  conversation to get to 7.
• 10:04
  SPEAKER_0: I was a coding machine from day one in unofficially.
• 10:08
  Literally as an intern,
• 10:10
  the first thing I did on my second or third day was I just noticed that the org
• 10:17
  chart tool looked really...
• 10:18
  It was a handmade tool at Facebook and looked really,
• 10:22
  really janky.
• 10:23
  SPEAKER_0: [inaudible 00:10:24].
• 10:24
  SPEAKER_1: This is internal tool where you can look up the org thing?
• 10:29
  SPEAKER_0: Yeah.
• 10:30
  I feel most companies have their own if they're not using something like
• 10:33
  Workday these days,
• 10:33
  but a lot of them build it,
• 10:34
  right?
• 10:35
  SPEAKER_1: Clearly, Meta [inaudible 00:10:36].
• 10:36
  SPEAKER_0: Yeah, and I don't know why.
• 10:38
  SPEAKER_0: I mean, Facebook built all their own tools, so I guess I know why.
• 10:41
  But it was just weird,
• 10:42
  like it showed two levels,
• 10:44
  it did lots of scrolling.
• 10:45
  SPEAKER_0: It's really weird.
• 10:45
  SPEAKER_0: So, I'm just going to rewrite this.
• 10:47
  SPEAKER_0: And the vertical height was weird, so I just made it horizontal.
• 10:52
  So Mark Zuckerberg was on the left and his reports were to the right and their
• 10:56
  reports were vertically below them.
• 10:58
  SPEAKER_0: So, it was like a horizontal layout and I just shipped this.
• 11:02
  SPEAKER_0: I wrote it, my manager approved the PR.
• 11:04
  It was second,
• 11:05
  third day as an intern,
• 11:06
  shipped it,
• 11:07
  and didn't get permission or anything like that,
• 11:09
  just drove it out the door.
• 11:10
  Didn't even really know what I was doing socially or social norms wise at the time.
• 11:17
  SPEAKER_0: I just thought it was like, "Cool, let's do this.
• 11:18
  SPEAKER_0: Let's crank out this code."
• 11:20
  SPEAKER_0: And that was really well received and [inaudible 00:11:23].
• 11:23
  So you ship this on,
• 11:24
  I don't know,
• 11:24
  second or third day,
• 11:25
  same day landing,
• 11:26
  and then suddenly the internal tool that everyone at the company is using,
• 11:29
  it just completely changed the visual design?
• 11:33
  And people thought it was fantastic because the horizontal nature,
• 11:37
  which was actually just a judgment call because the page was too tall vertically.
• 11:43
  SPEAKER_0: If it was a vertical layout, it was just too much scrolling.
• 11:46
  SPEAKER_0: But people loved it because it flattened the hierarchy.
• 11:48
  It showed that anyone can have influence across at any and on a horizontal line,
• 11:55
  like an intern is at the same line as Mark Zuckerberg.
• 11:59
  SPEAKER_0: And people love this.
• 12:00
  SPEAKER_0: So, it reinforced this idea that, wow, this is just how things work here.
• 12:06
  I can write as much code as I want and ship it as fast as I want and we'll see what happens.
• 12:09
  SPEAKER_1: Wow.
• 12:10
  And then there's lots of adventures after that,
• 12:11
  which [inaudible 00:12:13].
• 12:12
  SPEAKER_1: And this is like 2009, how big was the company roughly?
• 12:16
  It was about 220-ish,
• 12:20
  I'd say,
• 12:20
  about 200 engineers and about 700 employees at the time.
• 12:24
  SPEAKER_0: So relatively small, yeah.
• 12:26
  SPEAKER_0: Still in downtown Palo Alto scattered across the couple offices.
• 12:30
  SPEAKER_0: Yeah.
• 12:31
  SPEAKER_0: Sorry, that was a bit of a side track there.
• 12:35
  SPEAKER_1: What a story.
• 12:35
  SPEAKER_1: I love it.
• 12:36
  We were talking about what I did on a day-to-day basis I think as a coding machine.
• 12:40
  SPEAKER_0: So, I called myself a product engineer.
• 12:43
  SPEAKER_0: I was always officially allocated to a product team.
• 12:46
  I started off on internal tools,
• 12:48
  which is a product team at Meta,
• 12:51
  and then I worked on Facebook groups.
• 12:54
  And the first version of Facebook workplace,
• 12:56
  I worked on News Feed,
• 12:58
  I worked on Facebook groups for schools and colleges.
• 13:02
  SPEAKER_0: I worked on Messenger for kids.
• 13:04
  SPEAKER_0: That was one of the last things I worked on.
• 13:07
  So I was always on a product team,
• 13:10
  but I was always doing a lot of refactoring and code cleanup on this,
• 13:15
  "Ontheside."
• 13:16
  So a typical day was,
• 13:17
  I would say,
• 13:19
  contributing as a E-5 senior engineer locally on the team that I was on.
• 13:26
  Someone who's working on a pretty small team of 5 to 10 engineers leading projects,
• 13:32
  helping break things down,
• 13:34
  helping mentor,
• 13:35
  and then that was like I would do that in 30% of my time and spend the rest of my
• 13:40
  time on this massive refactorings and cleanups and these code base wide efforts.
• 13:49
  SPEAKER_1: Wow.
• 13:49
  So in 30% of your time,
• 13:51
  you did the work that for an E-5 or E-6 that would spend,
• 13:55
  I don't know,
• 13:49
  60,
• 13:49
  70,
• 13:49
  80% of their time depending on how you look at it?
• 13:59
  SPEAKER_0: Yeah.
• 13:59
  SPEAKER_1: Wow.
• 13:59
  SPEAKER_1: Impressive.
• 14:00
  SPEAKER_0: Yeah.
• 14:01
  But that's also why the coding machine progression is hard because it's if
• 14:07
  I'm acting like that role on a team,
• 14:10
  it's like if I spent that 33% of the time multiplied it by three and I just did
• 14:17
  replace three senior engineers,
• 14:19
  that probably wouldn't be enough to justify getting to the E-7 plus level at Meta.
• 14:25
  So that had to be weighed in with this outsized other impact that I was having.
• 14:32
  SPEAKER_0: Yeah, I don't know if that makes sense.
• 14:36
  And the concept of archetypes,
• 14:38
  it's super interesting that first of all,
• 14:40
  someone championed you.
• 14:41
  SPEAKER_1: Was it your manager at the time?
• 14:43
  SPEAKER_0: It was actually my skip level manager.
• 14:46
  Your skip level,
• 14:47
  but your skip level champion saying,
• 14:48
  "All right,
• 14:49
  here we have this IC doing great work,
• 14:52
  but they don't really have an archetype.
• 14:54
  SPEAKER_1: And let's create one to recognize and help them grow professionally."
• 14:58
  But it's taking a step back,
• 14:59
  the concept of archetypes is very unique to Facebook or Meta these days.
• 15:05
  I haven't seen it at many other companies,
• 15:07
  but what I do see is there's a glass ceiling of people being stuck unless you
• 15:14
  become a tech lead or more of...
• 15:16
  Usually,
• 15:16
  it's a tech lead where you go on and you have an impact.
• 15:20
  You've now with your current business,
• 15:25
  you have window into how other companies operate and how they interview and
• 15:28
  how to get in.
• 15:29
  Why do you think this is,
• 15:30
  that Meta still seems to be one of the few companies that has come up with this
• 15:34
  concept of career progression for very senior software engineers on the IC
• 15:38
  track with archetypes?
• 15:40
  SPEAKER_0: Yeah, I think it's a good question.
• 15:42
  SPEAKER_0: I think it reflects the company's culture.
• 15:46
  SPEAKER_0: It's an engineering first culture.
• 15:48
  There was tremendously talented individual contributors that I worked
• 15:53
  with early on in their careers,
• 15:55
  and Meta wanted to support those people and break through that idea that you
• 16:01
  have to be a manager to progress,
• 16:04
  so they wanted these people to be empowered.
• 16:06
  SPEAKER_0: And it's not just the levels.
• 16:08
  I feel like nowadays the companies are so large,
• 16:11
  people are...
• 16:13
  Maybe the way that IBM felt to me early 10 years ago is maybe how people might
• 16:18
  start looking at the structures of these big tech companies now,
• 16:21
  they just have tens of thousands of employees.
• 16:23
  SPEAKER_0: It's a lot more rigid.
• 16:26
  But Meta really wanted to have these individual contributors be superstars
• 16:32
  that were socially influential to at the company.
• 16:36
  And you would see at all hands,
• 16:39
  you wouldn't just see a presentation from the vice president of a department,
• 16:43
  you would see a presentation from the highest ranking individual
• 16:48
  contributing engineer as well.
• 16:50
  SPEAKER_0: And they had this equal presence.
• 16:52
  But I raised a lot of,
• 16:53
  or flipped a lot of tables,
• 16:55
  you could say,
• 16:56
  about this at the time.
• 16:57
  Because I felt like there still wasn't even enough support of ICs at the
• 17:01
  senior levels.
• 17:03
  At the more junior levels of management,
• 17:07
  there was about parity more or less between a first level manager and a staff engineer,
• 17:14
  the M1 and E-6,
• 17:15
  they were roughly equal percentages normalized and stuff.
• 17:19
  SPEAKER_0: But once you got the direct-
• 17:21
  I'm sorry,
• 17:22
  just to click on that,
• 17:24
  when you say roughly equal,
• 17:25
  do you mean in terms of influence,
• 17:27
  scope,
• 17:28
  compensation,
• 17:30
  expectations,
• 17:31
  that kind of stuff?
• 17:32
  SPEAKER_0: Yeah, compensation, like scope, all that stuff was roughly equal.
• 17:37
  And the ratio of senior managers to junior managers was roughly the same as
• 17:44
  E-7 to E-6,
• 17:45
  which are the parallel levels.
• 17:49
  And then once you got beyond that to the director level,
• 17:53
  which was D-1 and E-8,
• 17:54
  there was 10 x more directors or something like that.
• 17:57
  SPEAKER_0: I mean, there was like 10 E-8s at-
• 18:00
  SPEAKER_1: Uh-huh, I see what you mean.
• 18:00
  SPEAKER_1: Yep.
• 18:01
  SPEAKER_0: ... the company at the time.
• 18:03
  SPEAKER_0: There was like 20, 30 directors.
• 18:04
  And then that's when I started asking the question,
• 18:06
  okay,
• 18:06
  so there's this great career progression for ICs compared to all other
• 18:10
  companies that I know,
• 18:13
  but there is still this limit.
• 18:16
  SPEAKER_0: And I was really pushing back on that.
• 18:18
  SPEAKER_0: I wanted the infinite limit.
• 18:20
  I wanted to know why there weren't equal numbers of directors and principal
• 18:25
  engineer or whatever this partner engineers,
• 18:28
  I don't know,
• 18:28
  whatever the titles got beyond that,
• 18:30
  there isn't really titles.
• 18:33
  So I pushed a lot on that,
• 18:34
  and at the time the reasons weren't necessarily super clear,
• 18:40
  so I was really pushing on it.
• 18:41
  And in retrospect zooming out,
• 18:42
  I really see that there's a limit that an IC can have in terms of impact as an IC
• 18:49
  without having this manager behaviors.
• 18:57
  Even if you write 10 x the code or a hundred x the code,
• 19:01
  you cannot replace 30,000 engineers with one engineer.
• 19:05
  SPEAKER_0: So, you can be a VP of engineering though overseeing 30,000 engineers.
• 19:11
  So,
• 19:11
  no matter how multiply it out,
• 19:14
  I don't know,
• 19:15
  it's like some anime analogies where these characters get super jacked up.
• 19:24
  They can only get so powerful though as an IC,
• 19:27
  and managing all of those...
• 19:29
  If you're the manager of 10 of these superstars,
• 19:32
  you still have more influence/impact over the direction of those people in
• 19:39
  the company.
• 19:41
  So,
• 19:42
  I still don't have the best articulation of that,
• 19:46
  but I do understand now more once you start getting to those extremely senior
• 19:51
  levels that you can't really get past a certain point as an IC,
• 19:55
  but yeah.
• 19:57
  Is it fair to say that for these really senior IC roles,
• 20:01
  the E-8s,
• 20:02
  the principal distinguished partner engineers.
• 20:06
  Tenure becomes really important because I would assume that it's really
• 20:09
  hard to come from the outside and have this outside influence.
• 20:13
  Whereas if you've been there for a long time,
• 20:14
  you've helped build some of these systems,
• 20:16
  you know them very deeply,
• 20:17
  you have a super strong network,
• 20:18
  it's just a lot easier to have that outside influence.
• 20:21
  SPEAKER_1: Is this a fair observation?
• 20:23
  It's an excellent observation in question and point and discussion topic,
• 20:28
  for sure.
• 20:29
  Because this was actually one of the most interesting tensions I started
• 20:34
  observing when I got to that level.
• 20:38
  Anyone who's an E-7 plus,
• 20:40
  there's a private group for those people to interact.
• 20:44
  SPEAKER_0: That's like back channel, maybe sink on thing, whatever.
• 20:49
  SPEAKER_0: [inaudible 00:20:50].
• 20:50
  SPEAKER_1: This is probably at most companies probably by the way.
• 20:53
  SPEAKER_0: Yeah.
• 20:54
  And there's a in-person event too that they were doing at the time,
• 20:58
  and there was basically two groups of people in this bucket.
• 21:01
  SPEAKER_0: There's the outside people and the inside people, and very different.
• 21:06
  SPEAKER_0: Those aren't the right words because it sounds like [inaudible 00:21:09].
• 21:09
  SPEAKER_1: Yeah, but I understand what you're saying.
• 21:12
  External people who joined Facebook externally at a high level,
• 21:15
  and people who grew there from a low level.
• 21:17
  SPEAKER_0: And there was definitely a tension.
• 21:19
  SPEAKER_0: Like the Facebook culture is so strong, the values are so strong.
• 21:23
  The people who rose internally were generally promoted extremely quickly,
• 21:28
  similar trajectories to me.
• 21:31
  SPEAKER_0: And they had a very Facebook ingrained behaviors.
• 21:38
  People who came externally at these really high levels,
• 21:41
  they brought different perspectives from their previous companies.
• 21:45
  So,
• 21:45
  a lot of extremely senior people from Google and Microsoft brought their perspectives,
• 21:51
  but they definitely did not immediately align to the Facebook values the
• 21:57
  same way the internal people had built up over the years.
• 22:01
  SPEAKER_0: So there was definitely this difference.
• 22:05
  And generally speaking,
• 22:07
  the people who joined externally had a harder time having impact at those levels.
• 22:14
  And not a lot of them were at Facebook for a very long time,
• 22:20
  and a lot of them left.
• 22:22
  I don't know if this has changed in recent years,
• 22:25
  I would probably expect it would've changed.
• 22:27
  But at the time when there was only about a hundred of people at these levels,
• 22:31
  there was definitely that difference.
• 22:36
  I think it's a testament to how strong Facebook's values were actually meant
• 22:41
  something and were not something on paper.
• 22:44
  SPEAKER_0: If you were move fast and break things was a value.
• 22:47
  And if you moved really,
• 22:48
  really fast and you broke something,
• 22:51
  you were almost...
• 22:55
  Rewarded is not the right word,
• 22:56
  but it was not a bad thing.
• 22:58
  You were not fired for breaking something because you were trying so hard to
• 23:02
  live the values.
• 23:02
  Or if you were making a really bold bet because be bold was a value,
• 23:07
  and you were really pushing the limits of what people thought or pushing
• 23:13
  people to think differently.
• 23:15
  Like that was really rewarded even if your idea wasn't great for the business
• 23:20
  at first and it never worked out.
• 23:23
  SPEAKER_0: So I think that's why that happened.
• 23:25
  I can't speak to today though what happened since then,
• 23:28
  if that's still the case.
• 23:30
  In the AI talent wars that are starting,
• 23:34
  there's a lot of extremely senior people in AI who are being brought in from
• 23:38
  external who have unique skill sets that are that irreplaceable.
• 23:44
  And there's a lot of stuff going on there that might change that,
• 23:46
  but yeah.
• 23:47
  So,
• 23:48
  I'm interested in your story because when you told me that in six years you
• 23:52
  went from intern to E-7,
• 23:54
  which means you went intern,
• 23:56
  which I don't know is E-0 or something,
• 23:59
  E-3,
• 23:59
  a new grad,
• 24:00
  E-4,
• 24:02
  E-5 senior,
• 24:03
  E-6 staff,
• 24:04
  E-7,
• 24:05
  I think that's like four...
• 24:06
  I counted four promotions,
• 24:07
  and increasingly difficult ones.
• 24:09
  Usually a career path like this would take at least a decade if you're on a
• 24:14
  strong trajectory.
• 24:15
  Can you talk about what happened,
• 24:16
  how you basically rolled through the ranks so quickly and what you learned
• 24:22
  doing this?
• 24:24
  I was known at the company,
• 24:25
  company-wise being like the single-handed most all-in on Facebook person
• 24:30
  that you could imagine.
• 24:32
  When Facebook rolled out public facebook.com email addresses for all users,
• 24:38
  I shut down all my other email accounts and only used that one.
• 24:41
  SPEAKER_0: It was just like, all really all-in.
• 24:44
  And people were just like,
• 24:45
  "You don't have a Gmail account,
• 24:46
  thisiscrazy."
• 24:48
  But anyways,
• 24:49
  in terms of the progression,
• 24:52
  so I would say one level at a time.
• 24:55
  I have not afraid of heights but if you're doing a hike or something and it's
• 25:01
  really steep cliffs.
• 25:03
  If you zoom out and look at the cliff,
• 25:06
  you're like,
• 25:06
  "Oh my god," no one could walk on that path or climb that cliff.
• 25:11
  SPEAKER_0: This is impossible.
• 25:12
  But if you go there,
• 25:14
  look down and just focus on the next step,
• 25:16
  you can get there and you might realize that it looks a little different than
• 25:20
  when you were zoomed out.
• 25:22
  So,
• 25:22
  I really took one step at a time and was focused just on what's the next level,
• 25:30
  what are the requirements,
• 25:30
  am I doing this yet,
• 25:32
  et cetera.
• 25:33
  And there's two key things though that I think worked really well for me and
• 25:40
  they're very different things.
• 25:43
  SPEAKER_0: But the first one was my manager relationship.
• 25:48
  I don't mean this,
• 25:51
  you should be friends with your manager or I was friends with my manager.
• 25:55
  SPEAKER_0: In fact, I was a very difficult person to manage.
• 25:57
  And I don't know my managers liked me that much just because I'm a little impulsive,
• 26:01
  and I'm very passionate when I'm doing my work,
• 26:03
  and it's a little hard to wrangle me sometimes when I'm really in there.
• 26:08
  What I mean by the manager relationship was there was a lot of trust and
• 26:12
  transparency both ways.
• 26:14
  So my managers knew who they were working with,
• 26:17
  and I trusted that they knew who I was and that we were on the same page about
• 26:25
  putting me in the right spot to work on things.
• 26:28
  SPEAKER_0: And it goes both ways.
• 26:30
  SPEAKER_0: So, I also felt like I could help my manager do well at their job.
• 26:35
  So,
• 26:36
  if maybe they're a little overloaded with junior people on the team,
• 26:39
  helping to mentor those people.
• 26:42
  And having that more trusting,
• 26:46
  supportive relationship both ways I think really helped progress.
• 26:50
  My manager could be really blunt with me like,
• 26:53
  "Michael,
• 26:54
  I'm getting feedback.
• 26:55
  SPEAKER_0: You're causing too many bugs on this area of the code.
• 26:59
  SPEAKER_0: It's bothering this team, that's going to hold you back.
• 27:02
  SPEAKER_0: Work on that and [inaudible 00:27:04]."
• 27:03
  SPEAKER_0: I'm like, not defensive.
• 27:05
  SPEAKER_0: And I'm like, "Got it, I will work on that, but I will not slow down."
• 27:07
  SPEAKER_0: And we got that relationship.
• 27:09
  That's really important for me because it really helped me just move through
• 27:14
  the system.
• 27:16
  On the second thing,
• 27:17
  completely unrelated,
• 27:18
  I always had a notepad of things I was doing that might be useful in
• 27:23
  performance reviews or in summarizing my work later on.
• 27:27
  So if things happened in the moment,
• 27:29
  I don't know,
• 27:32
  fixed a really weird bug that was emergency bug that was flagged at the
• 27:36
  company level and I jumped in and fixed it.
• 27:38
  SPEAKER_0: I might just put a note on my notepad that I did that.
• 27:40
  And then when it comes around to performance review time or one-on-ones with
• 27:44
  my manager,
• 27:45
  I could go over these things I was doing and bucket them and be like,
• 27:48
  "I'm doing a lot of work that a lot of these things are contributing to the
• 27:53
  requirements of the next level.
• 27:54
  SPEAKER_0: So what else am I missing to get to this level?"
• 27:57
  SPEAKER_0: And then working on those things.
• 27:59
  SPEAKER_0: So, that really helped me.
• 28:01
  It's funny that you say this because this concept went viral when Julia Evans
• 28:07
  blogged about,
• 28:08
  she called it a Brag Document in my book,
• 28:10
  the software engineers guidebook.
• 28:11
  It's one of the best advices I gave to people who are focused on the career.
• 28:16
  SPEAKER_1: I called it a work log.
• 28:17
  It doesn't matter what you call it,
• 28:18
  but I've gotten feedback from people saying,
• 28:20
  "Oh,
• 28:20
  I did this one thing," that they heard about from somewhere from online and it
• 28:25
  made such a big difference.
• 28:26
  And it's such a small thing,
• 28:27
  but especially when you're doing a lot of unrelated things or here and there.
• 28:31
  SPEAKER_1: By the way, as a manager, when I used to be a manager, I loved it.
• 28:34
  SPEAKER_1: I loved it when people just brought this to me because I was like, "oh."
• 28:38
  As a manager,
• 28:38
  I didn't know about 90% of this stuff that was not the main project,
• 28:43
  and it was just really pleasant surprise.
• 28:46
  I'm assuming your manager's view shifted on you as well and knowing,
• 28:51
  "Oh,
• 28:52
  you're doing all these other things," which sounds like Meta.
• 28:55
  SPEAKER_1: Appreciated this as well, Facebook.
• 28:57
  SPEAKER_0: Yeah, exactly that.
• 28:59
  SPEAKER_0: Yeah.
• 28:59
  SPEAKER_0: It's a little tied in.
• 29:01
  I said they were unrelated,
• 29:02
  but it is tied into the manager relationship because if your manager trusts
• 29:05
  you and knows how to interpret your words and you give them this document,
• 29:10
  they know how exactly how to turn that into the right thing and what that means
• 29:16
  for calibrations and performance reviews and stuff.
• 29:19
  SPEAKER_0: And so, I plus one to that for sure.
• 29:22
  And speaking about Facebook's engineering culture,
• 29:24
  back then it was called Facebook.
• 29:26
  SPEAKER_1: Now it's called Meta so I'm going sometimes back and forth.
• 29:29
  SPEAKER_1: But what was it like, and starting with that, can you share?
• 29:34
  Like we know Facebook had internal tools,
• 29:36
  but what kind of internal tools were they?
• 29:40
  SPEAKER_1: Imagine if someone went back in time, what would you describe to them?
• 29:43
  SPEAKER_1: What did you see?
• 29:45
  Yeah,
• 29:45
  this is a good question because I started off on the internal tools team.
• 29:47
  SPEAKER_1: [inaudible 00:29:49].
• 29:48
  SPEAKER_0: I was originally going to go to grad school and do a PhD.
• 29:53
  And I really wanted to work on social networking things,
• 30:00
  but for companies.
• 30:02
  And not just companies,
• 30:03
  just like work,
• 30:04
  because social networks were really connecting people in a way never seen personally.
• 30:09
  And I was like,
• 30:10
  that would be really cool if our work was also similarly connected if we could
• 30:16
  just as easily find co-workers of co-workers and message them and stuff like that.
• 30:21
  SPEAKER_0: Which now seems like an obvious, but at the time it was not obvious.
• 30:27
  SPEAKER_0: It was very different.
• 30:28
  So something that's really cool about Facebook at the time was that internal
• 30:35
  tools were products.
• 30:37
  So they were products just like user-facing tools built on the exact same
• 30:41
  code base deployed to an employee subset of Facebook rather than the public
• 30:50
  website obviously.
• 30:51
  SPEAKER_0: So, it was internal tools were seen as product building.
• 30:56
  SPEAKER_0: They weren't seen as these annoying things that the company has to do.
• 31:01
  A lot of tools were seen at the time,
• 31:03
  it's like no one wants to be on the tools team,
• 31:06
  it's the slow moving boring team.
• 31:08
  But at Meta,
• 31:09
  it was our Facebook at the time,
• 31:10
  it was probably the most exciting team because you could build product at a
• 31:16
  speed for...
• 31:17
  Building product for 7,
• 31:08
  8,
• 31:08
  900,000 people,
• 31:19
  it was very much faster than building product for at the time,
• 31:23
  I think it was like 200 million people publicly.
• 31:26
  SPEAKER_0: So that was really, really exciting.
• 31:30
  Meta still,
• 31:31
  I think to this day has a culture of building a lot of tools from internal tools
• 31:35
  from scratch.
• 31:39
  One of the main reasons is they built all their infrastructure from scratch.
• 31:42
  There was no cloud providers at the time,
• 31:44
  so they had all their own infrastructure.
• 31:47
  They even built their own servers and eventually hardware for servers themselves.
• 31:53
  So,
• 31:54
  that mentality resulted in really custom tools to work with all this custom
• 32:00
  stuff that really just didn't make sense outside of it.
• 32:02
  But yeah,
• 32:06
  I think because of that,
• 32:07
  because all these tools were treated like products though,
• 32:10
  there was definitely this engineering product ecosystem the whole company
• 32:16
  has had as a result.
• 32:18
  Even if you weren't working on these tools or you were using these tools,
• 32:22
  you really felt like Facebook was a technology company that was solving even
• 32:27
  itsinternalfinancesusinginternaltoolsand-
• 32:31
  SPEAKER_1: Wow.
• 32:32
  SPEAKER_0: ... site integrity.
• 32:33
  It really felt like no matter what job you had at Facebook,
• 32:36
  it was a product engineering company.
• 32:40
  SPEAKER_0: And I think that's still stuck through to this day too.
• 32:44
  So then just to be concrete,
• 32:46
  there was a custom version control,
• 32:50
  it wasn't GitHub or something like that.
• 32:52
  SPEAKER_1: Custom code review, custom build tool.
• 32:54
  I know Facebook also open sourced some of this or at least release some of this,
• 32:58
  like the build tool.
• 32:59
  SPEAKER_1: I think was it Buck and Fabricator?
• 33:02
  They also released for some,
• 33:03
  but everything was just completely custom or internally built to work
• 33:09
  however you want it to work.
• 33:11
  Yeah,
• 33:12
  I think maybe in terms of everything you said except for source control,
• 33:17
  they were using SVN as the backbone and then people used at the time.
• 33:23
  I can't remember if they used Git first,
• 33:25
  but they basically jumped to using Mercurial locally on people's machines
• 33:31
  to manage their local branches.
• 33:33
  SPEAKER_0: But then you would always merge using SVN to the main branches and repos.
• 33:41
  And the reason they chose Mercurial was because it was a lot easier to work
• 33:47
  with the open source team to hack into it things that they wanted.
• 33:52
  SPEAKER_0: So it was still the same ethos, even though it was not everything.
• 33:56
  SPEAKER_0: The Git people were not as open to making some of those changes.
• 34:03
  SPEAKER_0: Yeah, building all those tools, the code review tools.
• 34:06
  SPEAKER_0: A lot of these tools have seeded, like the companies of their own now.
• 34:13
  SPEAKER_0: I think Statsig and Sentry, and I think there's some others.
• 34:18
  SPEAKER_1: So Statsig is experimentation, right?
• 34:19
  SPEAKER_1: Or it was seeded by the experimentation platform?
• 34:24
  SPEAKER_0: Yeah, Gatekeeper internally.
• 34:24
  SPEAKER_0: I mean, these are not like people [inaudible 00:34:29].
• 34:28
  SPEAKER_1: No, no, no, what I want but... I also know Graphite.
• 34:31
  They're a team,
• 34:33
  five Meta founders who are bringing the concept of stacked diffs to production.
• 34:38
  Again,
• 34:40
  inspired by this,
• 34:41
  it feels like there's a lot of teams who saw Facebook has something really
• 34:45
  cool that worked really well at a large company,
• 34:48
  but it just doesn't exist outside.
• 34:49
  And now they're productionizing it,
• 34:51
  and it seems like there's demand for it.
• 34:55
  Mark Zuckerberg said in,
• 34:55
  it's like a Y Combinator Startup School talk,
• 34:59
  I think it was 2009 or 2010,
• 35:02
  he said that he wanted Facebook to be the place that was known for training
• 35:07
  people how to build really good product and training people to potentially
• 35:13
  be really good founders.
• 35:15
  And I thought it was a bit weird at the time because...
• 35:17
  He also didn't talk that much publicly at the time.
• 35:19
  And I was like,
• 35:20
  "Oh,
• 35:20
  but what about training people to be really good engineers at Facebook to
• 35:24
  stayatFacebookforever?"
• 35:25
  And I guess it was the audience,
• 35:27
  it was a startup audience,
• 35:28
  but it got me thinking a little bit.
• 35:30
  And when you zoom out and look,
• 35:31
  it's like Facebook was exceptionally good at taking passionate,
• 35:36
  ambitious young engineers who don't have much experience,
• 35:40
  and building extremely strong abilities for them to execute and deliver.
• 35:46
  And we're seeing a lot of those products now being really robust,
• 35:50
  strong products that are coming out and helping people be on Facebook now too.
• 36:00
  One question I have,
• 36:00
  is it true that there was an internal tool just for booking meeting rooms and
• 36:06
  that it was awesome as well?
• 36:08
  SPEAKER_1: Someone told me.
• 36:12
  SPEAKER_0: Yeah, yeah, there was.
• 36:12
  My intern project when I started was also,
• 36:16
  it was a meeting scheduling tool because it was like,
• 36:20
  you just wanted to click on three people,
• 36:22
  find a meeting for them or find a time they can meet,
• 36:25
  find a meeting room,
• 36:25
  done.
• 36:26
  SPEAKER_0: So, I actually built that tool, yeah.
• 36:29
  SPEAKER_0: But that was-
• 36:31
  When I first heard this,
• 36:31
  I just had such a hard time believing because Facebook was already a lot
• 36:34
  bigger that A,
• 36:36
  someone would build this,
• 36:37
  B people would also use it,
• 36:39
  and C,
• 36:39
  there will be a team maintaining it.
• 36:41
  And somehow that all happened,
• 36:43
  because it sounded like anyone could build this in any company.
• 36:45
  SPEAKER_1: But maintenance would be a pain, the funding.
• 36:49
  Most companies would have a lot of politics in terms of like,
• 36:51
  "Okay,
• 36:51
  we need staffing for this team,
• 36:53
  we need budget,
• 36:54
  we need to justify.
• 36:54
  And in the end,
• 36:55
  and a few years later,
• 36:56
  "Oh,
• 36:56
  let'sbuyavendorsolutionorsomething."
• 37:00
  So it really surprised me that even for a relatively small but really
• 37:03
  important thing,
• 37:04
  it kind of happened there.
• 37:11
  SPEAKER_0: Yeah.
• 37:11
  When you talk about it now,
• 37:12
  it sounds like a...
• 37:13
  I don't know,
• 37:15
  Wild West place but...
• 37:17
  Or there's no problems,
• 37:20
  there's these things are just...
• 37:21
  It's like the best place for engineers,
• 37:25
  but there are those challenges too when you start to grow and scale.
• 37:28
  It's one of the reasons why as Facebook got bigger,
• 37:31
  I was less...
• 37:35
  That was not my thing,
• 37:36
  you could say.
• 37:36
  SPEAKER_0: The meeting rooms for example, right?
• 37:39
  So that was one of 40 tools that I worked on when I started was the tool to help
• 37:43
  people find and schedule meetings.
• 37:49
  SPEAKER_0: It was just a very quick, I don't know, fast thing.
• 37:52
  SPEAKER_0: It worked, did 90% of what was needed.
• 37:55
  SPEAKER_0: It was that vibe.
• 37:58
  And then as the company got physically bigger and physically had way more
• 38:01
  meeting rooms and you start diving to the next level of product problems,
• 38:06
  what if people book the room and don't show up?
• 38:10
  SPEAKER_0: Do you free the room?
• 38:11
  SPEAKER_0: What if someone else wants to use it?
• 38:12
  What if someone's sitting there and not doing a meeting,
• 38:15
  but they're just sitting there for fun?
• 38:18
  There's all those problems that you start...
• 38:20
  Just like I was saying,
• 38:20
  product problems even though they're internal.
• 38:26
  So now you start getting into,
• 38:26
  "Okay,
• 38:26
  doweneedtohavesensorsintheroomstoknowifsomeone'sactuallythere?"
• 38:30
  And then if we need to buy those sensors,
• 38:31
  we do need a budget for it,
• 38:33
  and then we need to install them and we need to install them across 20
• 38:36
  different offices.
• 38:37
  And then some offices have restricted floors for external people and you
• 38:43
  can't put them in those ones.
• 38:44
  SPEAKER_0: And you do end up getting into all those details too.
• 38:48
  SPEAKER_0: And I really love the product details.
• 38:50
  I think it's like,
• 38:51
  I love going to every single detail of a product and really ironing those out.
• 38:55
  But I do not like the budgeting side or the convincing other people that this
• 38:59
  is worth the budget side.
• 39:00
  SPEAKER_0: I'm just like, "Give me the keyboard and let me code."
• 39:04
  SPEAKER_0: That's me.
• 39:07
  SPEAKER_1: Yeah.
• 39:07
  So you did a lot of coding,
• 39:07
  but you were a pretty early employee at Facebook relatively.
• 39:12
  SPEAKER_1: Did you get to work with Mark Zuckerberg?
• 39:14
  Because I know a lot of early employees do have either collaboration or some
• 39:19
  stories of interacting or working with him.
• 39:23
  SPEAKER_1: What is yours?
• 39:26
  SPEAKER_0: Yeah, for sure.
• 39:26
  So yeah,
• 39:27
  I'm actually friends with him because I was there early and I did cross paths a
• 39:32
  lot and got to know him a little bit over the years.
• 39:35
  I mean,
• 39:37
  I can tell the story of how we became Facebook friends,
• 39:39
  that's kind of a thing.
• 39:42
  SPEAKER_1: Oh, okay.
• 39:42
  SPEAKER_1: All right.
• 39:43
  SPEAKER_0: I started as an intern.
• 39:44
  SPEAKER_0: I serve as an intern, like it's all these interns from school.
• 39:47
  Even though Mark was two years older than everyone or something,
• 39:51
  he was a college dropout.
• 39:54
  It wasn't like he was industry veteran,
• 39:58
  but he was well known and all the interns friend request him on their first day
• 40:06
  and hopefully he'll accept it.
• 40:10
  And it was interesting,
• 40:11
  the way that we became friends was he used to play Risk the board game.
• 40:18
  I think it was once a week people would get together,
• 40:21
  it was like four or five people and he would show up often and I really liked Risk.
• 40:27
  I would always play Risk as a kid for whatever reason,
• 40:29
  I don't even know why.
• 40:31
  And one day we're down to the final three people and it was me,
• 40:35
  him and I was losing.
• 40:38
  Mark was in second place,
• 40:40
  the leader [inaudible 00:40:43] was taking over.
• 40:43
  So I did a really,
• 40:44
  let's say,
• 40:48
  delicate strategic maneuver where I made an alliance with him to share
• 40:52
  resources to all go after the first place person.
• 40:56
  SPEAKER_0: So he went all in against the first place person.
• 40:59
  SPEAKER_0: My turn's next, I did not go all in against the first place person.
• 41:03
  SPEAKER_0: I took over Mark's remaining resources.
• 41:06
  SPEAKER_0: He just dwindled going after the first place person.
• 41:10
  SPEAKER_0: And he accepted my friend request very shortly that evening, I think.
• 41:20
  It's weird because I basically backstabbed him in the game really bad
• 41:28
  blatantly into his face.
• 41:31
  SPEAKER_0: But it's the game, that's what Risk is, and it's a strategy game.
• 41:33
  And I think that he appreciated the strategy that I had that it made him feel
• 41:40
  more like he could trust me even though I backstabbed him,
• 41:42
  he knows where my strategic thinking is coming from.
• 41:46
  That's something I've seen with Mark over the years with in product reviews,
• 41:50
  both myself doing product reviews and also product reviews with other people.
• 41:54
  SPEAKER_0: He asks a lot of questions.
• 41:56
  SPEAKER_0: He's very detail focused.
• 41:58
  SPEAKER_0: He really dives into trying to figure out the exact-
• 42:03
  Can you tell us what the product review is,
• 42:03
  because I don't think it's as common outside of Facebook.
• 42:08
  Yeah,
• 42:08
  it was called the Zuck Review at the time because he was the only person doing them.
• 42:11
  There was maybe 10,
• 42:11
  20 product teams of 5 to 10 people at the time roughly.
• 42:19
  And once a week they would do a Zuck Review,
• 42:22
  like a 15-minute presentation kind of thing.
• 42:25
  Usually,
• 42:25
  the PM or the eng manager would do them and sometimes members of the team would
• 42:31
  join if it was relevant.
• 42:33
  SPEAKER_0: It's like a doctor's office.
• 42:35
  SPEAKER_0: There's a waiting room of the teams, because they're so short.
• 42:39
  And one team would go in,
• 42:41
  do their Zuck Review,
• 42:41
  they would come out and you would see they got good news or bad news and you
• 42:45
  would see the expression on their face.
• 42:50
  Zuck would just give lots of feedback and steer the direction,
• 42:53
  and they're really short so it's very efficient feedback.
• 42:57
  SPEAKER_0: Yeah, those were so interesting.
• 43:01
  The more junior PMs would really stress about these because it was their
• 43:04
  FaceTime with the leader of the company.
• 43:06
  And I was just saying about the Risk story,
• 43:11
  Mark is judging constantly,
• 43:14
  does he trust you?
• 43:15
  SPEAKER_0: Does he trust your judgment?
• 43:17
  SPEAKER_0: Trying to read the situation.
• 43:18
  So it was a little stressful for junior PMs going in there and they would really...
• 43:23
  Sometimes the night before be trying to think of all different angles and
• 43:26
  they're like,
• 43:26
  "I got this backup slide in case Mark asked this question,
• 43:29
  andthisotherbackupslideincasethisone."
• 43:31
  And then you'd go into the review and you would interrupt in the first slide
• 43:35
  and just go into a completely different direction.
• 43:39
  SPEAKER_0: You would see that sometimes in the nicest way possible.
• 43:42
  SPEAKER_0: He is very not mean, not aggressive person.
• 43:46
  At the end of the day,
• 43:47
  it was about everyone delivering really good product and wanting to build
• 43:52
  the best product,
• 43:53
  and really trying all the hardest they could as individuals to get there and
• 43:59
  not sugar-coating anything,
• 44:01
  not patting on the back for no reason.
• 44:04
  We just all got to be really honest about what this product looks like and how
• 44:07
  do we make it better.
• 44:08
  SPEAKER_0: And yeah.
• 44:13
  You mentioned you try to work hard to build the best product,
• 44:16
  but you also wrote tons of code.
• 44:18
  So I'm going to assume you must have cost some outages because it just happens
• 44:21
  and move fast and break things.
• 44:23
  SPEAKER_1: What was your worst outage or worst bug?
• 44:28
  SPEAKER_0: There was this product that I don't know what the inspiration was.
• 44:30
  I didn't come up with a product idea,
• 44:32
  but we wanted to allow people to upload a photo template on top of their
• 44:39
  profile picture.
• 44:40
  So nowadays we see that on LinkedIn all the time with open to work or hiring,
• 44:45
  that kind of thing.
• 44:46
  This didn't really exist at the time,
• 44:48
  and they wanted to let people at Facebook do that.
• 44:50
  It would help increase engagement and people who haven't changed their
• 44:54
  profile photos for five years might use this to spruce it up a bit or something.
• 45:01
  So I built this prototype of this thing really fast in a week or something
• 45:09
  after it was handed to me.
• 45:11
  And then unfortunately,
• 45:12
  there was some terrorist attacks in Paris and a lot of people were expressing
• 45:19
  support on social media for the people who are impacted and trying to hope for
• 45:27
  good recovery.
• 45:28
  And using this new system,
• 45:32
  we released a template of the French flag that people could overlay over top
• 45:36
  of their profile picture to show support.
• 45:41
  What we didn't expect was a hundred million people to do that within a day or two,
• 45:45
  literally.
• 45:47
  SPEAKER_1: wow.
• 45:47
  SPEAKER_0: And this was a prototype.
• 45:48
  SPEAKER_0: So this is an interesting thing.
• 45:50
  I basically,
• 45:53
  the way that I built this is you have a template object,
• 45:56
  a photo object,
• 45:58
  all these things are federated across tens of thousands of databases.
• 46:01
  SPEAKER_0: So each object has a home database.
• 46:06
  So we need some way to tell this relationship between profile object,
• 46:11
  photo object and...
• 46:13
  Or sorry,
• 46:14
  template object.
• 46:15
  SPEAKER_0: So there's an edge concept to connect the two.
• 46:20
  And you can have a one-way edge that just goes from one to the other or a two-way
• 46:24
  edge where you go in both directions.
• 46:25
  You can go in both directions starting at the profile to the template or
• 46:29
  starting at the template to the profile.
• 46:32
  So sounds simple,
• 46:34
  sounds like systems design 101...
• 46:35
  It's systems design interview that at Formation we're helping people with
• 46:40
  all the time.
• 46:41
  SPEAKER_0: It's a classic problem.
• 46:47
  And then the system would just fetch these two objects and combine them or something,
• 46:48
  right?
• 46:49
  SPEAKER_0: Yeah.
• 46:49
  And typically at Facebook,
• 46:49
  most things start with the profile,
• 46:52
  like a person logs in and that's the root of all queries so to speak.
• 46:57
  SPEAKER_0: You start with the profile, then you get the profiles friends.
• 46:59
  It kind of edges,
• 47:00
  it's like a graph and you're jumping in at the profile and then crawling
• 47:06
  through the graph from there.
• 47:07
  And if the profiles are all federated across tens of thousands of databases,
• 47:11
  you can crawl out through a nicely distributed system without overloading
• 47:18
  any particular node.
• 47:20
  But when you have this template,
• 47:23
  which is an object in the system just like a profile,
• 47:26
  if you have an edge from that to the profiles and there's a hundred million of
• 47:30
  those written in a day,
• 47:32
  you have a hundred million edges going on that one node,
• 47:35
  one database that that [inaudible 00:47:39] single French flag template is
• 47:39
  stored on being written constantly.
• 47:41
  And rights are a lot more expensive in a system in at least Facebook system.
• 47:49
  And that node is one machine or virtual machine or something behind it,
• 47:51
  basically?
• 47:52
  SPEAKER_0: Yeah.
• 47:53
  SPEAKER_0: Exactly, yeah.
• 47:54
  SPEAKER_0: I mean, there's a lots of other objects on that machine too.
• 47:57
  There could be other profiles,
• 47:58
  it could be groups,
• 47:59
  there could be notes,
• 48:00
  there could be events,
• 48:01
  all kinds of stuff on that machine.
• 48:02
  SPEAKER_0: It's just the federation is generally not by object type.
• 48:09
  SPEAKER_0: So that machine basically melted.
• 48:12
  SPEAKER_0: I don't know, physically, I didn't hear if it may have physically melted.
• 48:17
  SPEAKER_0: No, probably not.
• 48:19
  So that machine crashed and then A,
• 48:22
  if...
• 48:23
  I'm trying to grab my friends,
• 48:24
  and my friends on that machine like that fails.
• 48:26
  SPEAKER_0: So, it causes cascading effects if that kind of thing can happen.
• 48:30
  SPEAKER_1: Wow.
• 48:32
  SPEAKER_0: Yeah, so it's interesting.
• 48:32
  SPEAKER_0: Now the interesting thing is it's a really bad bug.
• 48:35
  It's interesting because at Formation,
• 48:36
  we're helping people with systems design all the time.
• 48:38
  It's one of the most common things we're helping people prepare for in interviews.
• 48:43
  And when you talk about these things,
• 48:45
  this sounds like the most basic systems design system.
• 48:49
  You have federation is a common,
• 48:53
  and partitionings are very common topics.
• 48:56
  And this sounds like an obvious thing,
• 48:58
  like why would you potentially have a two-way edge on a system where one of the
• 49:06
  nodes might be super heavy?
• 49:08
  This is not an exact example,
• 49:11
  but Twitter had a lot of outages with the fail whales back then because they
• 49:16
  had a system that was similarly vulnerable when Justin Bieber tweeted or something.
• 49:22
  SPEAKER_0: They had a similar system where that was-
• 49:25
  SPEAKER_1: Yeah, they had the fan out problem, right?
• 49:27
  SPEAKER_0: Yeah.
• 49:27
  SPEAKER_0: So similar, yeah.
• 49:27
  SPEAKER_0: I mean, these are known problems.
• 49:30
  And it's interesting because I see so many people preparing when they're
• 49:35
  thinking about systems design interviews and stuff.
• 49:36
  They're thinking about the textbook answers for all these different things
• 49:41
  that can happen.
• 49:42
  And this is just one really simple situation that talking through why,
• 49:46
  like why did I make this two-way?
• 49:48
  SPEAKER_0: Because was I ignorant or was it not?
• 49:51
  SPEAKER_0: And the reason I did it was that this was a prototype.
• 49:55
  And when you're productionizing something,
• 49:57
  you need to put a lot of logging in for performance,
• 50:01
  for collecting metrics,
• 50:04
  for analyzing,
• 50:05
  like is this working,
• 50:06
  is it not?
• 50:06
  SPEAKER_0: All these behavioral metrics, all these things that you would do.
• 50:09
  You have to put a lot of time and effort into thinking about what are those things?
• 50:14
  Where do we log them,
• 50:16
  how do we log them,
• 50:17
  how do I explain to the PM how to do it?
• 50:20
  And if we just use the raw data model itself to store some of that,
• 50:23
  and we keep that two-way edge,
• 50:25
  it gives us more information.
• 50:26
  It lets us just count the number of edges as the number of people who have used this.
• 50:31
  SPEAKER_0: It's simpler for a prototype and we don't have to build all this stuff.
• 50:36
  SPEAKER_0: If we don't ship it, I don't want to rip out all this logging.
• 50:38
  I don't want to have wasted time explaining to the PM how the logging is working,
• 50:43
  those kinds of things.
• 50:45
  SPEAKER_1: Yeah.
• 50:45
  SPEAKER_1: Totally, yeah.
• 50:46
  SPEAKER_0: Not simpler, reasonable.
• 50:46
  SPEAKER_0: That's why the decision was made.
• 50:47
  SPEAKER_0: And then the second this machine melted, we knew exactly what to do.
• 50:50
  SPEAKER_0: And we [inaudible 00:50:53] that.
• 50:53
  SPEAKER_1: So how did you solve this outage, right?
• 50:54
  Because now,
• 50:54
  you actually had a problem,
• 50:55
  which sounded like it's at a data level.
• 50:57
  SPEAKER_1: And usually with an outage, the best way is to roll back the change.
• 51:03
  SPEAKER_1: Clearly not an option here.
• 51:05
  How did you fix it so that actually people could start to use those profiles?
• 51:12
  SPEAKER_0: Yeah.
• 51:12
  SPEAKER_0: So this actually was beyond my area, the DB.
• 51:17
  SPEAKER_0: There's a DB infrastructure team that stepped in.
• 51:20
  SPEAKER_0: I removed the edge, so we made it a one way edge.
• 51:24
  Due to the success of this feature,
• 51:26
  we also productionized it and addressed those more logging and things like
• 51:31
  that that we lose by chopping off that edge.
• 51:34
  Yeah,
• 51:36
  so the infrastructure team handles the rebooting the databases,
• 51:41
  the replaying of SQL logs that might be necessary because of lost
• 51:44
  transactions and things like that.
• 51:46
  SPEAKER_0: They handle that level.
• 51:48
  As a product engineer,
• 51:49
  you need to be aware of what's going on and maybe send over a six-pack of beer or
• 51:56
  a cake or cookies to the people afterwards to build up the reputation or...
• 52:04
  I'm assuming it was not the first time for them,
• 52:05
  like someone adding an edge and overloading a few nodes.
• 52:09
  Let's just say that the site reliability team had one of the most well stocked bars,
• 52:17
  like alcohol bars in the entire company.
• 52:23
  SPEAKER_1: They just kept arriving.
• 52:23
  SPEAKER_1: Thank you for this.
• 52:24
  SPEAKER_1: Thank you for that.
• 52:26
  SPEAKER_0: Yeah.
• 52:26
  SPEAKER_0: I'm actually not joking.
• 52:28
  SPEAKER_0: It was extremely well stocked.
• 52:30
  SPEAKER_0: I don't know if that's still allowed or if they [inaudible 00:52:35].
• 52:34
  SPEAKER_0: Yeah.
• 52:37
  SPEAKER_1: Yeah.
• 52:38
  SPEAKER_1: No, but lots of favors there.
• 52:39
  SPEAKER_1: That's a good one.
• 52:40
  So something interesting you told me before this one is that when it came to hiring,
• 52:48
  Facebook has this concept of the hiring committee,
• 52:50
  but it was also open door.
• 52:51
  So you could go in and you could shadow and sometimes...
• 52:55
  And you did shadow multiple times and then you participated.
• 52:58
  Can you explain to us the ones who...
• 53:00
  We were not at Facebook,
• 53:03
  what was this hiring committee meeting like?
• 53:07
  SPEAKER_1: I understand these were when they decided on the hire, no hire decisions.
• 53:13
  SPEAKER_1: How did they go?
• 53:14
  SPEAKER_1: What was it like to listen in, and what was it like to participate?
• 53:17
  And what does this tell about Facebook's hiring process for software engineers?
• 53:23
  SPEAKER_0: Yeah, maybe just a really quick recap of how you get to hiring committee.
• 53:25
  The Meta or Facebook interview process starts with a lightweight recruiter screen,
• 53:30
  get to know you,
• 53:33
  and then you'll jump into a coding interview.
• 53:36
  SPEAKER_0: It's called the technical screen.
• 53:38
  SPEAKER_0: It's like a 45-minute interview, usually two questions.
• 53:42
  If everything goes well,
• 53:43
  you'll go to the on-site where for most people,
• 53:47
  you'll do two more coding interviews,
• 53:49
  a behavioral interview and a systems design interview.
• 53:52
  SPEAKER_0: There's some subtle variations there you might do more-
• 53:57
  SPEAKER_1: They used have cool names, right?
• 53:57
  SPEAKER_1: These used to have cool names, I remember they were Jedi-
• 54:00
  SPEAKER_0: Jedi, Pirate, and Ninja.
• 54:04
  SPEAKER_0: And then, this is [inaudible 00:54:09].
• 54:09
  Ninja was a coding,
• 54:10
  Pirate with system design,
• 54:11
  and Jedi was the soft skills,
• 54:13
  behavioral.
• 54:14
  SPEAKER_0: Yeah, behavioral.
• 54:14
  SPEAKER_0: At the time it was half behavioral, half coding still.
• 54:15
  SPEAKER_0: They really wanted the coding signal.
• 54:17
  SPEAKER_0: And now I think it's all behavioral, but yeah.
• 54:23
  Because this is what we do at Formation all the time,
• 54:26
  two critically interesting things about Meta's interviews,
• 54:30
  specifically the technical ones that really throw people off to are the,
• 54:35
  they're whiteboarding style and there's almost no small talk.
• 54:39
  SPEAKER_0: And these two things can throw people off in the technical interviews.
• 54:44
  A lot of the times,
• 54:45
  you have 45 minutes,
• 54:46
  you're expected to code up solutions to two problems.
• 54:50
  So you jump in and sometimes the interviewer will just be like,
• 54:53
  "Hi,
• 54:53
  my name is blah,
• 54:54
  nice to meet you.
• 54:55
  SPEAKER_0: Here's a problem."
• 54:56
  And it can throw people off a bit because some people are...
• 54:59
  Well,
• 54:59
  everyone's nervous.
• 55:00
  I don't know anyone who's not nervous in interviews,
• 55:02
  but it can be a bit like you want to settle in,
• 55:05
  build a rapport.
• 55:07
  SPEAKER_0: Maybe if I screw up the question, they'll like me.
• 55:08
  SPEAKER_0: No, not for the coding interviews.
• 55:10
  SPEAKER_0: You just got to write the code.
• 55:12
  SPEAKER_0: So that's why one thing.
• 55:13
  SPEAKER_0: And then the second thing is whiteboarding style.
• 55:16
  SPEAKER_0: Now that things are remote, I don't even think you can compile the code.
• 55:19
  SPEAKER_0: They either block it or they tell you not to.
• 55:21
  So you can't just guess and check and enter to run your code to see if it works.
• 55:26
  They want to see you walking through a clear and clean thought process,
• 55:31
  problem solving process for how you're approaching the problem,
• 55:35
  how you're solving the problem.
• 55:36
  SPEAKER_0: They want to see your code being cleaned.
• 55:38
  They don't care if it has some minor syntax issues or you forgot a built-in
• 55:43
  functions signature and you just guessed it for the interview type thing.
• 55:49
  SPEAKER_0: Anyways, those are two things that are specific to their interviews.
• 55:55
  But anyway,
• 55:55
  you do your on-site,
• 55:56
  you do your technicals,
• 55:57
  your behavioral,
• 55:57
  your systems design,
• 55:58
  or maybe two systems design.
• 56:01
  And then if everything goes fairly well or better,
• 56:07
  then you'll go to hiring committee.
• 56:14
  There'll be a quick debrief with the people,
• 56:16
  so...
• 56:17
  With the interviewers,
• 56:18
  I mean.
• 56:18
  SPEAKER_0: So the interviewers will quickly sync sometimes over email even.
• 56:23
  SPEAKER_0: Were there any red flags?
• 56:24
  SPEAKER_0: Were there any strong no's?
• 56:26
  SPEAKER_0: You might just end the process there.
• 56:29
  So usually,
• 56:30
  teleformation all the time if you do a Meta interview and you don't hear back
• 56:35
  really quickly,
• 56:36
  then it's usually a good sign you were not quickly [inaudible 00:56:42]1.
• 56:40
  SPEAKER_0: You at least have a chance.
• 56:42
  So people get nervous when they're like,
• 56:43
  "Oh no,
• 56:43
  Ididn'thearbackinaweek."
• 56:44
  SPEAKER_0: And that's not a bad thing.
• 56:45
  SPEAKER_0: It's an okay thing or maybe even a good thing.
• 56:49
  SPEAKER_0: So anyways, you'll end up in hiring committee.
• 56:51
  At the time they met when I was there in the mid-2010s,
• 56:57
  they were meeting twice a week I think?
• 57:01
  And it was open,
• 57:03
  and this was really...
• 57:04
  I don't even know if it was officially open or I just went.
• 57:09
  SPEAKER_0: It wasn't publicly open, but it wasn't closed.
• 57:13
  There was a public calendar for engineering recruiting and it was on there.
• 57:19
  And all the recruiters who were representing a candidate to try to get
• 57:25
  approval for hiring would bring their candidate there with a packet.
• 57:28
  So,
• 57:29
  people were going in and out and I showed up and I was just started going to all them.
• 57:36
  SPEAKER_0: And I would have to figure out where they were.
• 57:38
  SPEAKER_0: And it was like a secret handshake, not figuring out what door to go to.
• 57:44
  Sometimes it would move last minute and figure out where it was sometimes.
• 57:49
  SPEAKER_0: But if you found it, you walk in and sit down at the table.
• 57:51
  SPEAKER_0: So what happens in hiring committee is a recruiter will present a packet.
• 57:57
  SPEAKER_0: This is actually pretty interesting.
• 57:59
  Very few engineers talk about this topic,
• 58:01
  by the way,
• 58:02
  of the hiring committee.
• 58:03
  Recruiters definitely would be able to,
• 58:06
  but I haven't seen a lot of them talking about it,
• 58:08
  so I won't give any secret sauce away but just the general process.
• 58:13
  A recruiter presents a packet to a quorum of leaders,
• 58:20
  and there had to be at least three director or VP engineering people there at
• 58:24
  the time.
• 58:25
  I don't think,
• 58:26
  it's not a legal constitutional number,
• 58:29
  but that was just they wanted a quorum of decision makers.
• 58:35
  SPEAKER_0: And recruiter would present the packet to them.
• 58:38
  SPEAKER_0: A packet has a bunch of information.
• 58:41
  SPEAKER_0: It has all your interview feedback.
• 58:47
  This is super,
• 58:47
  super interesting,
• 58:47
  but it has the interview history of the interviewers themselves.
• 58:52
  SPEAKER_0: It has the histogram of the interviewers feed of past ratings.
• 58:58
  SPEAKER_0: So is this interviewer super binary?
• 59:00
  SPEAKER_0: Their either strong yes, strong no, historically.
• 59:03
  SPEAKER_0: Are they lean yes or lean no.
• 59:05
  SPEAKER_0: So it helps the reviewers kind of calibrate, interpret people's feedback.
• 59:11
  SPEAKER_1: Feedback, nice.
• 59:11
  Are they a really experienced interviewer or are they a brand interviewer?
• 59:15
  It would have the questions they asked,
• 59:17
  it would have how many times they asked those questions.
• 59:19
  Did this interviewer ask this question for the first time,
• 59:22
  or have they asked it 150 times?
• 59:26
  It has a lot of information in the packet that was presented to the directors,
• 59:31
  and the directors would read it over,
• 59:34
  look for anything,
• 59:35
  any flags.
• 59:36
  Usually at this point,
• 59:37
  the default would be that someone would more than likely be hired.
• 59:41
  SPEAKER_0: So they're looking more for flags or I think inconsistencies.
• 59:48
  I can give tons of examples of what those are,
• 59:49
  which are...
• 59:49
  This is one of the things at Formation,
• 59:52
  one of the reasons why people work with us is getting to some of these nuances
• 59:57
  and stuff are pretty important,
• 59:59
  but might take hours.
• 1:00:00
  So I won't go into all of them,
• 1:00:02
  but yeah,
• 1:00:03
  they're looking for flags and different things.
• 1:00:05
  SPEAKER_0: The most common thing that they're looking for is level.
• 1:00:09
  I reiterate the fairness and consistency that Meta wants to have in its
• 1:00:14
  hiring process,
• 1:00:16
  and they want to bring in someone at appropriate level where they're an
• 1:00:21
  appropriate level consistent with other people who perform or have similar
• 1:00:26
  scopes of responsibility.
• 1:00:27
  SPEAKER_0: So they want to make sure that your level's right.
• 1:00:30
  And that was probably the biggest thing that was discussed was like,
• 1:00:35
  is this person a senior?
• 1:00:37
  SPEAKER_0: Are they mid-level?
• 1:00:38
  SPEAKER_0: If they're a mid-level, do they have too much experience?
• 1:00:43
  Is there career progression a little bit slower than we wanted for mid-level?
• 1:00:45
  And maybe we'll reject or are they actually at the senior level and this is simple?
• 1:00:52
  Or did they interview and appear with some of the traits that a senior
• 1:01:00
  engineer might have,
• 1:01:01
  but they're clearly mid-level from their background and they want to make
• 1:01:05
  the call on where does the offer end up?
• 1:01:07
  SPEAKER_0: Stuff like that.
• 1:01:08
  SPEAKER_0: They're working through all those things.
• 1:01:12
  SPEAKER_0: So, why did I go is maybe an interesting question.
• 1:01:16
  SPEAKER_0: This sounds boring maybe, but I was very interested in-
• 1:01:20
  SPEAKER_1: Oh, let's get into it.
• 1:01:21
  ...
• 1:01:21
  making sure that...
• 1:01:21
  I was paranoid that that Facebook would lower its hiring bar as they scaled.
• 1:01:28
  Even now,
• 1:01:29
  I don't really like saying lowering the bar because I think that...
• 1:01:32
  Especially working at Formation and the secret sauce,
• 1:01:37
  I think to interviewing and finding a job is finding the right job now.
• 1:01:42
  So,
• 1:01:43
  just because you didn't "Meet the bar for Meta" does not mean that you're a bad
• 1:01:47
  engineer and not hireable,
• 1:01:49
  it just means that there wasn't some alignment.
• 1:01:53
  SPEAKER_0: It might be them, it might be you.
• 1:01:58
  There's too many ways to overthink it,
• 1:02:00
  but finding that right role is way more important than just checking out the
• 1:02:04
  boxes and passing the interview and feeling great that you got the offer.
• 1:02:07
  SPEAKER_0: Yeah.
• 1:02:10
  In the time though,
• 1:02:13
  at the time I was just paranoid,
• 1:02:15
  I don't understand how the company can grow 10x and not lower the bar accidentally,
• 1:02:22
  or I'm paranoid about this.
• 1:02:24
  SPEAKER_0: Like I want an extra check.
• 1:02:25
  So I was going to these things to see who's being hired and seeing if I had any flags.
• 1:02:29
  And I would chip in occasionally with my 2 cents,
• 1:02:32
  and sometimes it had some impact,
• 1:02:37
  sometimes it didn't.
• 1:02:39
  SPEAKER_1: Well, I guess, look at how things work out now.
• 1:02:41
  It turns out those sessions were pretty helpful now doing what you're doing
• 1:02:45
  at Formation,
• 1:02:46
  right?
• 1:02:48
  SPEAKER_0: Yeah.
• 1:02:48
  SPEAKER_0: Really uniquely [inaudible 01:02:50].
• 1:02:49
  Coaching people,
• 1:02:49
  helping people,
• 1:02:49
  the fact that you were in the room is a big difference.
• 1:02:55
  SPEAKER_0: Yeah.
• 1:02:59
  Even if it's just to reassure them if they don't get selected,
• 1:03:04
  but they thought they did really well.
• 1:03:08
  You can say things with words,
• 1:03:10
  but I really have the experience there to help people get that it might not be
• 1:03:19
  that they failed or that it's just could be a lot of reasons.
• 1:03:25
  So in terms of interview processes,
• 1:03:26
  you are intimately familiar with Meta's,
• 1:03:28
  but of course because of what you do at Formation,
• 1:03:31
  you're familiar with a lot of other big techs.
• 1:03:33
  What company do you think has the most similar recruitment process for
• 1:03:39
  software engineers as Meta has these days,
• 1:03:42
  if any?
• 1:03:46
  SPEAKER_0: Yeah.
• 1:03:47
  So,
• 1:03:47
  I think the lineage is pretty interesting because the interview processes
• 1:03:54
  are fairly similar to a lot of the top companies and there are some
• 1:03:59
  differences here and there,
• 1:04:00
  and some things have changed a bit over the years.
• 1:04:02
  SPEAKER_0: But generally speaking, LeetCode wasn't a thing.
• 1:04:05
  SPEAKER_0: It didn't exist as a word until 2012 I think was when they were founded.
• 1:04:10
  SPEAKER_0: And the problems haven't changed.
• 1:04:13
  The interview styles haven't changed since before LeetCode,
• 1:04:16
  since after LeetCode.
• 1:04:18
  SPEAKER_0: It's pretty consistent style that has been around.
• 1:04:22
  And Facebook didn't invent these things,
• 1:04:26
  they kind of borrowed a little bit from Yahoo.
• 1:04:28
  Yahoo was one of the big "Successful companies" that was on the shrinking
• 1:04:34
  side at that time when Facebook was growing.
• 1:04:37
  SPEAKER_1: Oh, interesting.
• 1:04:38
  SPEAKER_0: So they borrowed some things from Yahoo, from Microsoft, from Google.
• 1:04:42
  SPEAKER_0: Google was the preeminent place to work at the time too.
• 1:04:45
  SPEAKER_0: And Google borrowed from other companies.
• 1:04:47
  And now we see a lot of the AI companies like OpenAI has a fairly Facebook ish
• 1:04:58
  process and vibe.
• 1:05:03
  Yeah,
• 1:05:03
  we're seeing this,
• 1:05:05
  and ultimately I think the reason why this is the case is that these interviews,
• 1:05:11
  especially the...
• 1:05:12
  We call them the LeetCode interview,
• 1:05:14
  but at Formation,
• 1:05:16
  we don't like pigeonholing the naming to being about doing a problem,
• 1:05:25
  clicking a button and something saying you passed or that kind of thing.
• 1:05:29
  The point of these interviews is to test stack agnostic and language
• 1:05:33
  agnostic problem-solving skills.
• 1:05:35
  SPEAKER_0: And that has always been the point.
• 1:05:36
  That's like what engineers do,
• 1:05:38
  and these interviews haven't changed as stacks have evolved.
• 1:05:44
  They're really trying to take away any biases that might come from being a
• 1:05:47
  JavaScript programmer or a C++ programmer.
• 1:05:50
  SPEAKER_0: Just focus on the core logic, the problem-solving process.
• 1:05:57
  SPEAKER_0: And that's what they're trying to do and that's what engineers do.
• 1:05:59
  So,
• 1:06:00
  it's very understandable that many truly technology and engineering
• 1:06:04
  companies use these processes.
• 1:06:07
  And the companies that don't are generally smaller companies because they
• 1:06:13
  do actually want expertise in specific stacks or specific experiences or
• 1:06:18
  non-tech companies that don't really have the same ethos.
• 1:06:26
  SPEAKER_0: But yeah, so we actually see it pretty variations of these.
• 1:06:31
  SPEAKER_0: Some companies do some more practical non-DSNA style stuff, but... Yeah.
• 1:06:40
  Yeah,
• 1:06:40
  it sounds like it's smart if you want to get into one of these companies,
• 1:06:43
  just go through,
• 1:06:44
  prepare,
• 1:06:45
  because it's useful everywhere,
• 1:06:46
  right?
• 1:06:48
  SPEAKER_1: You master it once and it is useful probably for years or even beyond.
• 1:06:54
  SPEAKER_0: Yeah, yeah.
• 1:06:54
  One of the interesting things that Formation that we've seen that shocked us
• 1:06:57
  is we actually see people coming back after...
• 1:07:01
  They come to Formation,
• 1:07:02
  we'll help them with prepare for these data structure algorithms,
• 1:07:06
  systems design interviews.
• 1:07:07
  They get a job and then two years later they come back and do it again and pay us again.
• 1:07:12
  SPEAKER_0: It was one of those surprising things.
• 1:07:13
  I mean,
• 1:07:15
  from a business point of view,
• 1:07:16
  that'sthedreamisyouwanttohaveservicethat-
• 1:07:19
  SPEAKER_1: Yeah, recurring customers.
• 1:07:23
  Yeah,
• 1:07:23
  service that's useful to people for their entire lives is way more valuable
• 1:07:27
  as a business than a service that's useful at once.
• 1:07:30
  SPEAKER_0: So, that's actually pleasantly surprising to us.
• 1:07:35
  I mean,
• 1:07:37
  we're going on a little bit of tangent about interviewing now,
• 1:07:39
  but I think the interview preparation process,
• 1:07:44
  I draw a lot of analogies to like a personal trainer like you want to get in shape.
• 1:07:49
  SPEAKER_0: You might've been in amazing shape 10 years ago.
• 1:07:52
  You remember what it was like to have the habits to be in shape,
• 1:07:56
  to eat well,
• 1:07:57
  to do your exercise.
• 1:07:58
  And you can do it again,
• 1:08:00
  but you're not in shape now and you need to maybe get in shape again.
• 1:08:04
  And whether you've got a gym membership or a personal trainer or you just do it yourself,
• 1:08:11
  there is this recurring pattern of getting ready for these interview processes.
• 1:08:19
  There's infinitely many stacks and combinations of frameworks that an
• 1:08:22
  engineer can work with.
• 1:08:23
  And when you work with your frameworks and your stacks for years and years,
• 1:08:29
  it's like you're working out,
• 1:08:31
  but you're only doing leg exercises,
• 1:08:32
  so maybe your legs are really strong and your upper body's not as strong
• 1:08:36
  because you haven't been doing that.
• 1:08:37
  And you have these different transformations that you make just from your
• 1:08:41
  day-to-day job.
• 1:08:42
  So,
• 1:08:43
  check when you're making a career transition to a new job,
• 1:08:46
  like getting back in shape again,
• 1:08:47
  it makes a lot of sense.
• 1:08:51
  SPEAKER_1: Yeah.
• 1:08:51
  And I know people complain about how interviewing is different than the job,
• 1:08:53
  but it is,
• 1:08:54
  and it's how it is.
• 1:08:56
  SPEAKER_1: I think you can either fight it or accept it.
• 1:09:01
  But speaking of this topic,
• 1:09:04
  what advice would we have for more experience software engineers who are
• 1:09:07
  interviewing right now,
• 1:09:08
  just in the sense that we know that the job market is a bit tougher than before.
• 1:09:13
  What kind of tactics have you seen or strategies that help people land offers
• 1:09:19
  at some of these better known companies?
• 1:09:22
  SPEAKER_1: What are things that are more important than before?
• 1:09:26
  SPEAKER_0: Yeah, it's definitely changed over time.
• 1:09:30
  I mean,
• 1:09:30
  when I joined at a school,
• 1:09:31
  it was like,
• 1:09:33
  I think there were so few engineers.
• 1:09:35
  I know people who are trying to get H-1B Visa's were just being handed them
• 1:09:40
  becausetheyhadsuchasurplus-
• 1:09:44
  SPEAKER_1: Wow.
• 1:09:44
  SPEAKER_1: Completely different time.
• 1:09:45
  And now there's 20x oversubscription of people trying to get tech jobs in the
• 1:09:49
  US and specifically as the one example.
• 1:09:52
  There was a different vibe back then I would say,
• 1:09:55
  but now it is more competitive.
• 1:09:57
  There's a new team matching process that we're seeing a lot of companies have.
• 1:10:01
  SPEAKER_0: Google's had this for a long time.
• 1:10:04
  SPEAKER_1: Oh, interesting.
• 1:10:05
  SPEAKER_0: Google's had this process for a long time.
• 1:10:06
  SPEAKER_0: Meta's more recently added the team matching process.
• 1:10:08
  We're seeing a couple of other companies,
• 1:10:10
  smaller companies,
• 1:10:11
  but not do this.
• 1:10:14
  It's like you pass the onsite,
• 1:10:16
  but now there's another...
• 1:10:17
  You're in limbo for a while until you actually match to get the offer.
• 1:10:23
  Yeah,
• 1:10:24
  so we are seeing that competition be distributed across a couple of
• 1:10:30
  different places,
• 1:10:33
  adding a couple extra layers,
• 1:10:35
  a little bit more filtering on the front end.
• 1:10:36
  We're seeing more companies give online assessments right off the bat
• 1:10:42
  instead of jumping into the technical interview to filter more people there.
• 1:10:47
  So we see a lot of people who have 10 years of experience,
• 1:10:50
  they've never used the tool like CodeSignal or HackerRank and they fail one,
• 1:10:55
  and they're like,
• 1:10:55
  "What is going on?
• 1:10:56
  SPEAKER_0: I've never seen this tool before."
• 1:10:57
  SPEAKER_0: And we're like, "You have to practice that."
• 1:11:00
  SPEAKER_0: Yeah.
• 1:11:00
  Definitely seeing that the competition's definitely distributed across
• 1:11:03
  each level of this process,
• 1:11:05
  for sure.
• 1:11:08
  SPEAKER_1: Nice.
• 1:11:08
  SPEAKER_1: All right.
• 1:11:09
  SPEAKER_1: Can we wrap up with some rapid questions?
• 1:11:13
  SPEAKER_0: Yeah, sounds good.
• 1:11:14
  SPEAKER_1: All right.
• 1:11:14
  SPEAKER_1: So let's wrap the questions.
• 1:11:15
  I just ask the question and then you just fire out whatever comes to your mind
• 1:11:21
  or the one that you can recall.
• 1:11:24
  I've been wanting to really ask this,
• 1:11:25
  how many diffs did you close or code in your most productive year when you were
• 1:11:29
  a coding machine?
• 1:11:30
  SPEAKER_1: Did you keep track of that?
• 1:11:35
  SPEAKER_0: I did keep track of it because I was in a competition.
• 1:11:36
  SPEAKER_0: At first, I was competing with other people.
• 1:11:38
  Sorry,
• 1:11:38
  it was rapid,
• 1:11:39
  but I was competing with other people to try to get more code than them.
• 1:11:42
  Once I surpassed everyone,
• 1:11:44
  Iwascompetingwiththisreleasebotthatreleaselike-
• 1:11:49
  SPEAKER_1: The release bot, no.
• 1:11:52
  SPEAKER_0: Yeah.
• 1:11:52
  And then after that,
• 1:11:54
  I think I stopped counting,
• 1:11:57
  but it was a lot in the thousands.
• 1:11:59
  SPEAKER_0: I honestly write more code now though even after leaving Facebook app.
• 1:12:02
  SPEAKER_0: At Formation, I'm writing a lot of code, so yeah.
• 1:12:06
  SPEAKER_1: Yeah.
• 1:12:06
  SPEAKER_1: So multiple diffs per day basically.
• 1:12:09
  SPEAKER_0: For sure.
• 1:12:09
  Yeah,
• 1:12:09
  it's probably around 5 to 10,000 a year in the double digits per day,
• 1:12:13
  but that doesn't really mean that much.
• 1:12:16
  SPEAKER_0: Some are small, some are gigantic.
• 1:12:17
  SPEAKER_0: It's hard to... Yeah.
• 1:12:21
  SPEAKER_1: Yeah.
• 1:12:21
  But your code actually did meaningful work that was part of why you got promoted,
• 1:12:24
  the difference.
• 1:12:27
  SPEAKER_0: Yeah.
• 1:12:27
  SPEAKER_0: I mean-
• 1:12:28
  Which language did you write the most code in when you were at Meta,
• 1:12:29
  and which one do you write the most code in now?
• 1:12:34
  SPEAKER_0: So then Hack, which PHP evolved into Hack which is their version of it.
• 1:12:40
  SPEAKER_0: Hardly any JavaScript.
• 1:12:41
  And then after leaving,
• 1:12:42
  I had to learn JavaScript as almost like a junior,
• 1:12:47
  and react and get up to speed on that.
• 1:12:49
  SPEAKER_0: And now I'm a very all JavaScript, yeah.
• 1:12:53
  SPEAKER_0: Front end and back end.
• 1:12:56
  SPEAKER_1: And which one's your favorite programming language and what's the reason?
• 1:12:59
  So it's grown on me,
• 1:13:01
  but I would say TypeScript specifically,
• 1:13:04
  which I don't really call it a language,
• 1:13:06
  but...
• 1:13:07
  So,
• 1:13:07
  at Meta I wrote...
• 1:13:10
  I sound like telling the long time ago stories as a grandparent or something,
• 1:13:16
  but I wrote all my code in VIM with no IDE,
• 1:13:21
  no syntax completion.
• 1:13:22
  SPEAKER_0: All I had was syntax coloring.
• 1:13:24
  SPEAKER_0: I almost had the code base in my head to code.
• 1:13:28
  SPEAKER_1: Wow.
• 1:13:30
  And now that I've slowly warmed up to TypeScript,
• 1:13:32
  I actually really like it because it helps me catch a lot.
• 1:13:34
  It hardly any guardrails,
• 1:13:37
  I still can code almost as fast as I would with no types,
• 1:13:40
  but it actually catches bugs,
• 1:13:43
  helps catch things,
• 1:13:45
  also helps me be a little less lazy if I'm just going to throw some parameters
• 1:13:50
  onto a function to think a little more about do I type?
• 1:13:54
  SPEAKER_0: How do I type these, do I move them into a structure, that type of thing.
• 1:13:58
  SPEAKER_0: But yeah.
• 1:14:00
  SPEAKER_1: And what's a book that you read that had an impact on you?
• 1:14:04
  SPEAKER_0: I think it was called Histories.
• 1:14:06
  SPEAKER_0: It's by Herodotus, the Greek historian.
• 1:14:11
  And the reason why it really sticks with me is a lot of the stories in there
• 1:14:16
  don't seem to make a lot of sense factually,
• 1:14:19
  and I'm a very factual,
• 1:14:20
  literal person.
• 1:14:22
  And when I read this,
• 1:14:22
  I'm like,
• 1:14:22
  "Wow,
• 1:14:23
  these stories don't seem to be real,
• 1:14:25
  buteveryone'strustingthese."
• 1:14:27
  And it showed me that storytelling is really important to try to communicate
• 1:14:31
  things that are happening,
• 1:14:32
  whether you're literally saying things that happened or not.
• 1:14:37
  The point is that to influence and to share and to be human,
• 1:14:42
  we have to be able to communicate stories.
• 1:14:45
  SPEAKER_0: So that showed me the power of storytelling, historical events.
• 1:14:53
  SPEAKER_1: Thanks very much for being on the podcast.
• 1:14:53
  SPEAKER_1: This was very, very informative and lots of fun.
• 1:14:59
  SPEAKER_0: Yeah, thanks for having me.
• 1:14:59
  It was a great time to revisit some old stories and yeah,
• 1:15:02
  had a great time.
• 1:15:03
  SPEAKER_0: Thanks.
• 1:15:05
  Thanks to Michael for sharing his thinking stories and how a coding machine
• 1:15:07
  thinks about things.
• 1:15:08
  SPEAKER_1: You can find Michael on LinkedIn and Reddit with links in the show notes.
• 1:15:12
  For more details about Meta's engineering culture and how hiring processes
• 1:15:15
  work at various tech companies,
• 1:15:17
  see deep dives from the pragmatic engineer also linked to the show notes.
• 1:15:21
  If you will enjoy this podcast,
• 1:15:22
  please do subscribe on your favorite podcast platform and on YouTube,
• 1:15:25
  it helps the podcast a lot.
• 1:15:27
  SPEAKER_1: Thanks, and see you in the next one.