Kesean Woodhouse

Engineer working at the intersection of systems and people.

I build technical systems and help the people who use them — backend, integrations, automation, and the customer-facing work that connects the two.

About

I've spent the last decade somewhere between technical and people-facing work. I started in tech retail at the Microsoft Flagship store in NYC — teaching customers, training employees, running events, and quietly handling a lot of the operational glue that kept the place running. I went to Flatiron School, joined the Microsoft LEAP apprenticeship, and worked as a software engineer at Microsoft and Stripe.

What I've built has mostly lived in the connective tissue: data pipelines that make systems talk to each other, automation tools that take manual work off engineers' plates, and integrations that help customers actually use the products they've bought. I'm comfortable across C#, Python, JavaScript/TypeScript, and modern cloud platforms. I work best when I can see how the technical pieces map to the humans on the other end.

I'm currently completing a deliberate career break — using the time to upskill in modern AI tooling, ship a full-stack production project, and figure out what kind of work I actually want next. Looking for technical roles where I can stay close to both the systems and the people, at companies that value sustainable pace as much as good engineering. You can also see this story compressed onto a single page → Resume

Projects

A full-stack developer support chatbot, built as a learning project.

After two years away from full-time engineering work, I needed a project that would force me back into the rhythm of shipping — and force me to learn the modern infrastructure I hadn't touched yet. Golem is the result.

It's a developer support chatbot that answers technical questions in a structured format: summary, root cause, debug steps, and relevant docs. Users sign in, ask questions, and get streaming responses. History is saved per-user. Under the hood, it pulls relevant documentation chunks from a vector store before generating each response.

The tech: Python/Flask backend, React/TypeScript frontend, Clerk for auth, Convex for the database, Qdrant for vector search, Voyage AI for embeddings, Redis for rate limiting, Railway for the API, Vercel for the frontend, GitHub Actions for CI, and Playwright for end-to-end tests.

I built this with a lot of help from Claude Code. Most of the production-grade pieces — vector search, RAG pipeline, rate limiting infrastructure — were tools I'd never used before. I shipped first to validate that the whole system works end-to-end, and I'm now going back through each layer to actually understand the why behind every choice. The point of the project wasn't to ship a product; it was to get my hands back on modern tooling and prove to myself I could still navigate complex integrations after time away.