Software engineer working across machine learning and data — the model and the system that has to serve it, since the second part is where most of them die. I like problems where failure is measurable: forecasts you can score, latency you can budget, benchmarks that test what they claim to test.
|
btc-alpha Low-latency crypto prediction for BTC/ETH/SOL/XRP futures. Rust core, Python training pipeline — research → paper → live. |
crucible RL environments for evaluating AI coding agents on bugfix, feature, refactor and performance tasks — reproducible workspaces, verified golden solutions. |
Also somewhere in the pile: a Bybit trading dashboard (Fastify + FastAPI + Next.js over 1-second market frames) and a portfolio recommender that explains itself in plain English.


