Hey, I'm
Staff Software Engineer building full-stack products and AI-powered systems.
I wanted a real number for what React Native costs, so I shipped a GPS speedometer in Expo and then rebuilt it in SwiftUI with the same algorithms, the same constants and the same colours. The native build is 27× smaller. The difference I actually noticed was somewhere else entirely.
I spent a day tuning settings to make a 27B model answer faster, and none of them worked. Switching to a 3B-active mixture-of-experts model made it 5–7× faster, and a 707-question check found it wasn't any worse.
Slim, distroless, alpine, or a compiled single binary — benchmarked on a dedicated amd64 host under cgroup limits. The base image turned out not to matter at all, and it took me two wrong results to find that out.
An offline GPS speedometer for iPhone. No account, no network, no analytics — the app has no networking code in it at all.
A browser multiplayer FPS in the style of Warzone's Resurgence mode — squads, respawns until the circle closes, 29 guns, and four movement styles. Three.js client, authoritative Node server.
An image CAPTCHA-solving API that returns a calibrated confidence score, so callers know when to trust the answer. One Go binary and Postgres in front of any OpenAI-compatible vision model.