Preston Brown

What I keep chasing is whatever is underneath the abstraction. I want to know what the compiler did, where the memory went, and what the operating system is doing while my program thinks it is the only thing running. The fastest way I have found to learn any of that is to build a worse version myself and then find out why mine is worse.

Most of that went into Pekoscript, a language I designed and wrote the compiler for. Type checker, static analysis, an LLVM backend, a language server. That is the work I would most want to be judged on.

What has changed recently is that I care a lot more about the gap between building something technically good and building something people actually need. I spent five and a half years on a hard technical problem, got it working, and then had to admit I had never seriously asked whether it mattered. That was expensive.

So now I try to do it the other way around. I am still figuring out which problems are worth years, and I would rather say that than pretend I have it sorted.

I wrestled through high school and train jiu-jitsu now, I spent years in Scouting leadership, and I listen to an unreasonable amount of yacht rock.

The short version is that I learned to build by building, and then spent five and a half years on Peko, turning web apps into native ones. I wrote the language and the compiler underneath it. It worked. I eventually decided the opportunity was not good enough to justify another five years, so I stopped.

What came out of that was a method rather than a company. Now I look for real work that real people do, try to understand it properly, and only then ask what software could do about it. That is what the loan processor research is.

ProxOS is the small side project running alongside all of that. It is not impressive and it is not trying to be. I am writing every line of it myself, which after specifying an engine and having it built for me was sort of the point.