Treblle hackathon, 24 hours

Runtime

An API monitoring platform, both halves, in a day.

Latency
p50 · p95 · p99
Throughput
requests per minute
Errors
4xx and 5xx by route
runtime-rust.vercel.app
The Runtime analytics dashboard

Treblle set the brief: build something useful for API monitoring. I built both ends. Requests pass through an Express proxy that logs them, scores each one for security problems, resolves the caller to a rough location, and rolls the results up into pre-computed metrics so the dashboard is reading summaries rather than recomputing on every load. The Next.js front end sits on top of that.

A deadline that short changes how you work. Aggregating metrics up front was the decision that paid off, since it meant the dashboard stayed fast without any caching work. The one I regret is hard-coding every threshold and time window to whatever looked reasonable at 3am, because nothing is configurable without a redeploy. SQLite was the right call for a hackathon and the first thing I would replace.

Front end
Next.js, TypeScript, SWR, Recharts
Back end
Node, Express, TypeScript
Data
Prisma, SQLite
Auth
JWT and OAuth
Built in
24 hours
Source
Frontend and backend on GitHub