The revised rules of engineering management with Will Larson

Gergely Orosz (0:00)

How has engineering management changed in the last few years? Will Larson is one of the best people to talk about this. He's been an engineering executive for the last 10 years at four different companies, wrote some of the most widely-read books on engineering leadership and staff plus levels, and I really appreciate how deep and thoughtful of a thinker he is. Today we cover how zero interest rates heavily influence a decade of management habits in the 2010s, many of these habits no longer helpful, how the role of engineers is changing, and why solving problems that the business cares about is becoming much more important than the way we thrive inside of companies. The path to becoming a CTO today, and why line managers often make better engineering executives than middle managers do, and many more. If you are an engineer or engineering manager and are wondering what your career path could look like with AI changing the industry, and how the path to CTO looks like today, this episode is for you.

This episode is presented by Turbobuffer, a ridiculously scalable, fast and cheap hybrid search engine built on top of object storage by an engineering team that I've really grown to like after spending time with them. The Turbobuffer engineering team is doing something really cool. They are completely redesigning their storage architecture from Perth's principles to make search faster, cheaper and more reliable at scale. The new system called Turbobuffer V3 is not yet in prod, but it already passes CI. Still, it's not nearly as performant as the current production version, V2, and so the T-Buff team is now Perth maxing query plans until they are as fast or faster than V2.

They're doing all of this in public, documenting how they increase performance all the way up to the V3 production launch. And it's so cool to follow along and see the different performance wins and how they did it. For example, we see how FullTech Search, the red line, started out as 129 times slower than the current system, but now it's only 4 times slower. And in the development log, we can also see that the team used things like batch scoring or batch reads to increase performance. A lot of these wins are basic foundational work right now, but the team will be sharing some very cool systems work over the next few weeks. But hang on, why is the team doing a floor rewrite? It's because search is changing, and so is TurboPuffer. TPUF started as a vector database, but it's no longer VectorDB. It has evolved into a high-performance database for search. The V3 discards the vector primary index, and all indexes will share the same columnar storage system.

Every query will run on vectorized query engines, which is something you don't usually see in search databases, but which will result in both very fast search and just generally a very fast database.

The TurboPuffer team is keeping a work log of their journey, and you can follow along at turbopuffer.com/v3. It's a blast to see how it's being built.

Gergely Orosz (2:32)

Will, welcome to the podcast.

Will Larson (2:34)

Thank you for having me. Excited to be here. Finally, chat in person.

Gergely Orosz (2:37)

Finally, in person because we know each other for so long.

Both started writing. We will talk about that writing as well, and also just hung out before, but this is nice. I want to jump into the beginning. How did you get interested in computers? Were you interested in computers to start with?

Will Larson (2:53)

From the time I was five, we had a Mac 2C or something like that. This was not black and white, not the very earliest one, but it was pretty basic. I think it had one thing that I understood how to use as a four and a half or five-year-old, which was KidPix.

Did you ever use C KidPix?

Gergely Orosz (3:09)

I know of it, but I never used it.

Will Larson (3:11)

It wasn't a super sophisticated thing. I don't even know if you could save files, but it was basically a graphical program for children, and you could make squares and circles and little graphics. That was kind of the introduction to me, and I remember using that computer a lot.

This is an era when there was really no internet to speak of, really no way to figure out how to do anything other than just banging on it, which is sort of the fundamental of writing software. In some ways, it's just banging until you figure out how to make it work. I think that was a great introduction, but to be honest, I wasn't doing a whole lot in software other than multi-user dungeons. Did you ever play a MUD growing up?