3 merged PRs across production LLMOps codebases at YC-backed startups (26k★ repos). I verify root cause before touching code, and ship fixes maintainers don't have to rewrite.
🔨 Right now: shipping fixes across LiteLLM, Agenta, and TraceRoot — router retry logic, schema validation gaps affecting 26+ models, tool-call ID sanitization, and 2 dependency CVEs.
| Repo | Scale | What I actually fixed |
|---|---|---|
| LiteLLM (LLM Gateway) | 26k★ | Audited 28 Databricks model configs, found 26 silently missing schema validation — fixed provider-wide, not patched one model. Also: Bedrock tool-call IDs breaking multi-turn conversations, Router discarding provider backoff signals, a live budget-bypass bug in spend tracking, 2 CVE patches. |
| Agenta (LLMOps, YC) | 4.2k★ | Breadcrumb routing bug + trace exception display fix |
| TraceRoot (Observability) | 341★ | Normalized error handling across 8 different API providers — survived 6 rounds of maintainer review before merge |
Synthetic Oracle — ML stock signal engine XGBoost + RF + 4 more models, 74% accuracy, 542 NSE stocks, 13-layer decision engine, SHAP explainability, Groq LLM analysis. Live · Code
CodeArena — Docker-sandboxed coding platform Judge0 execution, real-time submissions, AI code review via Groq/Llama. Live · Code
StartupOS — Ops automation dashboard GitHub API + Google Sheets + Slack webhooks, rule-based alert engine, multi-tenant JWT auth. Live · Code
Node.js · Express · MongoDB · React · Python · FastAPI · Docker · TypeScript · REST APIs · JWT · XGBoost
📩 Open to remote backend / AI infra roles.
