I design AI products and ship them to production.
I design AI products that feel human, and deliver them as production-quality code. I work on the software most designers avoid: data-center operations and consoles where a wrong click costs real money.
Selected work
Rebuilding the repair console for Meta's data centers
The operator tool for diagnosing and fixing hardware at scale. Faster repairs, and far fewer tickets that come back a second time.
Turning Microsoft Planner into an AI PM you can trust
An agent that drafts the plan and keeps it current, designed so people believe what it tells them.
Redesigning fulfillment operations at Amazon
Multi-Channel Fulfillment rebuilt for sellers: real-time tracking, flexible orders, mobile-first for global reach.
Cutting data access from ten days to forty-eight hours
Cornerstone, Amex's internal data-access portal, redesigned so the fast path is the safe one.
InfraBench: published specs, corrected for reality
Free, open planning tools for data-center people, built on public data with every constant shown.
Where I plug in
AI product design
Copilots, agent dashboards, operator consoles. Built so people trust them and actually enjoy using them.
AI enablement for design teams
Tooling and workflow set up inside your design org, so the whole team ships faster, not just me.
Design, delivered as code
I build what I design: working code your engineers can review and extend, not a spec to reinterpret.
Companies I've designed for
I make complex, technical products feel obvious, and I ship them.
I design with AI coding tools, delivering working code instead of static specs. I started as a brand designer in ed-tech, moved through enterprise finance to Amazon and then Meta, where I shipped 1,000+ production code changes solo on an internal AI platform. Now I take select consulting and freelance work.
Infrastructure, internal AI tooling, operator software. I go looking for that kind of complexity, then spend most of the project making it not feel that way to whoever has to use it.
I design it and I build it, so discovery-to-build takes weeks instead of quarters, and nothing gets lost between design and engineering.
I picked up the rigor at companies big enough to need it, and I've worked without it long enough to know which parts you can skip. Design system one month, scrappy MVP the next.



