Building Claude Code with Boris Cherny

**Gergely Orosz** (0:00)
What happens when you join one of the top AI labs in the world, and your first pull request gets rejected? Not because the code was bad, but because you wrote it by hand. This is exactly what happened to Boris Cherny when he joined Anthropic. Boris is the creator and engineering lead behind Claude Code. Before joining Anthropic, he spent 7 years at Meta, where he led Code Quality across Instagram, Facebook, WhatsApp and Messenger, and was one of the most prolific code authors and code reviewers at the company. In today's episode we cover How Claude Code went from a side project to one of the fastest growing developer tools, and the internal debate at Anthropic whether to release it at all. Boris' daily workflow of shipping 20-30 pull requests a day with zero handwritten code, and how code review works when AI writes everything. Why Boris believes we're living through a time as transformative as the printing press, and which engineering skills matter more now, and which ones do not. If you want to understand how one of the people closest to AI coding agents actually builds software today, and what that means for the rest of us engineers, this episode is for you. This episode is presented by StatSig, the unified platform for flags, analytics, experiments and more. Check out the show notes to learn more about them, and our other season sponsors, Sonar and WorkOS. How did you get into tech, software, engineering and coding in general?

**Boris Cherny** (1:14)
It starts a while back. I think there was kind of like two parallel paths that crossed. So when I was maybe 13 or something like this, I started selling my old Pokemon cards on eBay. And I realized that on eBay, you can actually write HTML. And I was looking at other people's Pokemon card listings, and I realized some of them have big colors and fonts and stuff like this. And then I discovered the blink tag.

**Gergely Orosz** (1:39)
The name was blink tag.

**Boris Cherny** (1:41)
And if I put the blink tag on it, I could sell my card for like 99 cents instead of 49 cents or whatever. So I kind of learned about HTML this way, then I got an HTML book and kind of learned about HTML. And then the second thing was this is also, I think sometime in middle school, we had these old TI-83 graphing calculators and we used them for math. And what I realized is I can get a better answer on the math test if I just program the answers to the math test into my calculator. And so I wrote these little programs. I just program the answers and then the test got harder. So then I had to program the solvers instead of the actual questions, because I didn't know what the coefficients and stuff would be ahead of time. And then the math got more advanced the next year. And so I had to drop down from basic to assembly to just make the program run a little bit faster.

**Gergely Orosz** (2:27)
Oh, so again, in high school, you dropped down to assembly.

**Boris Cherny** (2:30)
I think it's like middle school or high school, maybe like eighth or ninth graders. I don't know about this. Then the thing I realized is everyone in my class was starting to realize that I had the solver and they got jealous. So I bought this little serial cable so I can give it to them too. Then the next math test, everyone on the class just got A's and the teacher was like, what's going on? Eventually, she realized it. She was like, okay, you get away with it once and knock it off. But for me, it was very practical. In school, I studied economics. I should drop down to start startups. I never thought that coding would be a career at all. It was always very practical to me. Coding is a means to build things. And to make useful things. This startup, the first one was, I think it's like my friends and I were trying to get weed.
So we started this weed review startup. We made a website, we called different dispensaries, I think. And then we just tried to get weed samples so we could review it for them. And it actually blew up. And then I actually got more interested in, at the time, no one was testing this stuff.
And so I got into kind of the chemical testing, kind of chemical analysis. And then after this, I kind of did a bunch of other startups. And then I joined YC actually pretty early. And I was the first hire of this YC startup up in Palo Alto after.