Navigating from individual AI adoption to team-level transformation while scaling w/ Jakub Oleksy #268

Jerry Li (0:03)

This episode is brought to you by Sinch. Here's the thing about Sinch, you know them better than you realize. Think about it, last time you got a text confirming delivery, a call verifying it was really you logging in, and an email landing exactly when it's supposed to. You probably didn't think about it at all, and that's the point. Behind every one of those moments is Sinch. They're the trusted communication infrastructure that the AI error runs on. Enterprise will create messaging, voice, and email, with delivery, routing, compliance, and fraud prevention. Built in from day one, with more than 600 direct carrier connections, powering one trillion interactions a year.

That's why over 200,000 businesses worldwide rely on Sinch Network. When a product has to reach real people reliably at scale, that's who you build on. So the next call, text, and email your business sends is powered by Sinch. Check them out at sinch.com. That's sinch.com. Github is an incredibly complicated system. I can't go tell an agent, hey, just make Github faster. It's too complicated. You need to break these problems down. And a big part of breaking the problems down is actually understanding the system. And so we use AI tools a ton for just data analysis, performance analysis, and out of that, then come insights, then we could apply back to actually go fix the system. And sometimes the fixes, you know, you might spend three days analyzing a complicated situation and the fix itself might be a one-line, right? So the volume of code isn't really the measure. The outcome is the measure. And a lot of the investment that AI can help with is that analysis.

Patrick Gallagher (1:31)

Hello and welcome to The Engineering Leadership Podcast, brought to you by ELC, the engineering leadership community.

Jerry Li (1:38)

I'm Jerry Li, founder of ELC.

Patrick Gallagher (1:40)

And I'm Patrick Gallagher, and we're your hosts. Our show shares the most critical perspectives, habits and examples of great software engineering leaders to help evolve leadership in the tech industry.

Jerry Li (1:52)

Jakub, I'm really excited to have your show today. There's so many things I want to learn from you, especially the world we're in now. A lot of people are leaving and brazing Github all day long for many years, and it's a piece of infrastructure that always works. But a lot of people don't know what's happening behind the scenes, especially now. Essentially, every engineer becomes a manager, like running agents. So, we'll have to dive deeper into this conversation and learn from your perspective. So, maybe you can tell us more about Jorio and what you're focusing on at Github. Yeah. Thanks for having me. I'm really excited to be here and talk about this as well. It's a super exciting time at Github and the industry, and looking forward to the opportunity to talk about what's going on and what we're doing about it. My role is VP of Infrastructure and Enterprise at Github. I've actually been here for six years and seen a lot.

The last year and a half have been particularly exciting. A lot of my focus right now is squarely in the eyes of scale, performance, reliability. That's where a lot of the focus is. Obviously, a big part of my team is continuing to work on product features as we shift toward enterprises, but my personal focus is very much on scale and reliability as we face this quite unprecedented time of growth. Yeah. What does scale really means differently now compared to six years ago? Yeah, that's a good question. I think there's a few dimensions. So one is just the basics. There's more requests, more traffic, more use of the platform. But I think maybe a piece that goes underappreciated sometimes is the complexity of the system has actually gone up a lot. If you look back six years ago to what Github was, comparing it to what it is today, in terms of the breadth of features, the scope of product and what we actually provide to our customers and that could customers depend on every day, there's a lot. There's a lot of different pieces of the product. It isn't really only about Git. There's an entire ecosystem built on top of source control that we provide that our customers depend on. And so scale isn't just about sheer throughput of Git requests. It's actually the entire system working in connection to provide these services to our customers. Because Github was originally built for humans, but now we're living in a very different world and the rebuild is actually happening. What some of the first signals that indicates that whatever we have been planning for scaling is no longer viable. Yeah, you start out with an interesting point. Github was built for humans, obviously. Fundamentally, it's a collaboration platform. The interesting aspect of that, though, is that agents and how they work today, there really are very much behaving in a way a human would just a lot faster. And so a lot of the primitives that we have, and if you look at Github as a platform, it still works. It actually works very well, which is why we're seeing such a huge spike in usage, because those primitives that have always helped humans collaborate are now helping humans and agents work together and produce value. For us, if we sort of take a step back and look at the history of, let's say, the last year, really the critical pivot point came around November. When Opus 4.5 came out, that was the first moment where we really started to see a bigger bend in the curve when we looked at traffic in our system. We had been planning, as any enterprise does, as any engineering team does, for refactoring, growth, capacity management, all of those things are standard operations. But the pace at which we were doing that was based on the trajectory that we saw. Then again, when Opus 4.5 came out, I think engineers went into the holidays, that time period and started playing with these new models. All of a sudden, there's a lot of excitement. If you go back and you look at social media posts during that time, there was a lot of rumblings like, wow, these things are pretty good. Then folks I think came back from the holidays and started applying what they learned to work.