Full-stack engineer — backends, APIs and the automation that connects them, in production on Linux, macOS and Windows. I take a half-built system, a broken integration or an idea that only exists in your head and hand back something running, documented and yours to operate.

Backends and APIs in Python (FastAPI, Flask) and Node/TypeScript: auth, rate limits, idempotent handlers, retries with backoff, queues, Postgres and SQLite in WAL mode for concurrent writers. Data work: Playwright, Puppeteer and raw CDP for sites that fight automation, parsing, dedupe, resumable runs, scheduled delivery, structured logs so failures surface the same day. Integrations: payments, CRM, email, sheets, internal systems and APIs never meant to be automated, mapped from real traffic when undocumented. Systems: Linux, macOS and Windows, systemd units and timers, launchd and Task Scheduler, supervision, Docker, reversible deploys, monitoring and tested restores. Also React dashboards, OpenCV/YOLO/OCR, backtesting in Python, and LLM features that earn their place: retrieval, tool-calling agents with guardrails, multi-provider routing with failover and cost control.

Proof: 50+ model routing with ranked failover; tool schemas cut over 80% of their prompt cost while 1,100+ tools stay reachable; browser-session-to-API with file and shell tools; a multi-agent platform with a one-key kill switch.

Fixed scope, written plan, testable milestones, docs and handover. 3-14 days. US Eastern.

Send me the job that keeps breaking.