Engineering with the whole product in view.

I came to software through building independent products and learning what it takes to carry an idea into a working application. That means thinking past the screen: identity, data ownership, external services, deployment, monitoring, and what a person needs when something goes wrong.

My work spans frontend applications, APIs, databases, infrastructure, AI systems, and developer tooling. I’m especially interested in agentic systems, Model Context Protocol, and the engineering choices that make production AI useful: bounded actions, reviewable output, source trails, and clear handoffs to people.

How I think about building

I treat a model as one component in a larger system. Good products make the surrounding workflow understandable: what context went in, which tools ran, what changed, and where confidence or evidence is missing. The same principle applies outside AI—keep responsibilities clear, make important state visible, and design recovery alongside the happy path.

I like working across the stack because product decisions often cross technical boundaries. A good interaction may depend on a backend contract, a database choice, a deployment constraint, and a clear way to review the result.