Ship LLM apps
RAG pipelines, agentic workflows, evals — and the boring infra to keep them honest.
Twenty years building products for five million users. Now I run a one-person studio — designing, engineering, and shipping AI-native software on my own.
I treat shipping as the only signal that matters. Small batches, real users, fast feedback — and a refusal to scale code or process before the product earns it.
One week pinning down the actual problem, the audience, and the smallest version that proves it. No sprawling discovery decks — just sharper questions.
AI-augmented full-stack work — design, code, infrastructure — taken from blank page to production in weeks, not quarters. Boring stack where it should be, novel where it has to be.
Once real users are on it, every decision is grounded in data and direct contact. I keep the loop tight: ship, observe, decide what to cut.
Selected work
A few products that earned real users — from K-pop audition platforms to industrial IoT systems.
Capabilities
Twenty years across mobile, web, infra, and now AI. The list below is what I'll actually reach for on day one.
RAG pipelines, agentic workflows, evals — and the boring infra to keep them honest.
Mobile, web, backend, infrastructure — one person, one cohesive product.
AWS, GCP — boring where it counts, observable from day one.
From a fuzzy idea to a product real people use — without a team to hide behind.
// Toolkit
Get in touch
Tell me what you're trying to ship. I'll reply within a day — usually within an hour.