About

I've spent eight years designing products across AI, fashion, mobility, and EdTech. The core of that was at Snap, shipping end-to-end experiences for major retail clients, from early research through to production.

I love the point where design systems meet real user behaviour, getting the architecture right so the experience feels inevitable rather than designed. I do my best work in the room from day one, not handed a brief after the decisions are made.

Two questions kept following me. How do you make a model's output feel trustworthy rather than arbitrary? And how do you design for latency, uncertainty, and the first session before personalisation kicks in? At Snap they were the brief for Muse, an AI stylist I designed from 0 to 1.

Recently I set that brief for myself and shipped Roam, an AI trip planner, solo, from first research note to a live app tested by real users. What Muse and Roam both taught me: good AI product design isn't about the interface layer. It's about closing the gap between what the model knows and what the user expects, enough that they keep going. And with today's tooling, one designer can own that entire problem, research to deployed code.

When I'm not designing I'm usually experimenting with a new recipe, trying to keep my plants alive, exploring a new neighbourhood in Berlin, or watching something obscure on a streaming platform nobody I know has heard of.