latest blog post
AI Code Review Benchmark
August 3, 2025
I built a lightweight benchmark to test 5 major AI models on real pull request decisions from major open-source projects like Kubernetes and VS Code. Most models turned out to be "Yes Men" - approving 80-90% of PRs including problematic ones.
Read more →View all posts →
about me
hi! i recently graduated uc berkeley with a computer science degree, now working as a software engineer at atlassian.
connect with me:
projects
chibify
an app that converts your photo into your chibi version
aiart
a space to train art styles and generate art
autochapter
automatically chapter lecture videos, generate chapter summaries, and search for specific chapters
art
some of my art
honey
traffic