I build agentic workflows, observable LLM applications, and event-driven backends that move from promising demos to dependable production systems.
02 / Deployed systems
Production AI is an architecture problem. These systems make model behavior observable, keep slow work off request paths, and preserve control at every handoff.
SYSTEM / ONLINEAutomated executive workflows across email, meetings, and scheduling.
SYSTEM / ONLINEProcessed incoming business documents asynchronously without blocking the product experience.
SYSTEM / ONLINEMigrated the payment system from Stripe to PayPal while preserving customer access.
Third-party proof
Contributions across AI platforms and developer tools demonstrate how I debug unfamiliar systems, respond to review, and deliver durable fixes in public.
Enterprise MCP and AI platform
Built an end-to-end Outline knowledge connector with incremental sync, checkpoint management, 17 unit tests, and frontend integration. Also restored access to files attached in chat.
Connector shipped after multiple review rounds
Mac cleanup and optimization tool
Contributed fixes for installed-app cache filtering and log cleanup paths, with work credited by the maintainer in project release notes.
Maintainer-recognized production fixes
Open-source AI agent platform
Diagnosed and fixed EventSource memory leaks in the frontend API and improved contributor setup by standardizing Supabase CLI usage.
Improved stability and developer experience
Production AI · backend systems
How I work
I'm a software developer with 2+ years of experience across AI-powered products, scalable backend services, and full-stack applications. I work closely with product and frontend teams to turn ambiguous workflows into systems that can be operated, debugged, and extended with confidence.
Writing
Field notes on full-stack development, AI systems, and the engineering decisions behind production software.
