About
I run UX research at Suno, where people make music with AI. I joined in March 2024, when we were fewer than 20 people, built the UX research team (five people so far), and run it as an AI-first lab. Most of my time goes to deciding with the founders and product team what to build next, and checking whether it worked.
Before Suno I spent three and a half years at Spotify as a research scientist, working on the AI DJ and on music and podcast recommendations. Before that I did a PhD in computer science at Brown, building apps that helped thousands of people run experiments on their own sleep.
One study at Spotify changed how I think about recommendations. Most recommenders hand you a ranked list, so you never see how the music connects. We built TastePaths, a map of a genre anchored on artists you already listen to, and tested it with listeners. Their best discoveries were between genres or at the edges of them. And they wanted to steer: grow the map where they liked what they heard, and cut the parts they didn't.
That pattern shows up in everything I work on. I call it guided agency: people want the AI to do the hard part, but they also want to steer, and most products only get one of those right. The gap is usually small, the last 10% between what the AI made and what the person had in mind, but it's where people decide whether the result is theirs. I wrote more about where the idea came from in What is guided agency?
It's also why I trust what people notice more than what benchmarks say. A model can improve on every benchmark and still feel the same to the person using it.
Selected papers
Contact
If you want to talk about any of this, I'm on LinkedIn. My case studies are private, so ask there if you'd like to see them.
My team is hiring two senior UX researchers, one in New York and one in Los Angeles.