AI Engineer World's Fair 2026 — unofficial import demo

We haven't done a code review in two years. The last time I read every line of code in a PR was about six months ago. And we build a programming language with a runtime meant to replace V8. This is real engineering: compiler internals, runtime behavior, type systems, codegen, concurrency semantics, and FFIs across multiple languages. The thing that makes this possible is a technique we call "fight slop with slop" - every line of code is analyzed in depth by a sprawling toolchain of custom visualizers, linters, test snapshots and a whole bunch more. While the core language VM code has super high standards, a lot of these meta-tools are mostly vibe-coded. I'll dive deep into all the tactical things we've built, and how to adopt "fight slop with slop" in your own team