I build tools that turn messy, fragmented information into decisions people can act on.

I learned this at the Shahid Afridi Foundation in Pakistan, planning health outreach across communities where clinics ran on survival-mode logistics. The constraint wasn't expertise. It was that recording patient demographics and prescriptions at the point of care was impossible at that volume. So I rebuilt where the data came from, inferring regional drug usage from inventory that was already being counted while collecting demographic data separately in the field. Resources and clinicians could finally respond to actual local need rather than guesswork.

I've been working on versions of the same problem ever since: how to make decisions when the underlying information is fragmented, noisy, or missing entirely. Whether building systems to evaluate health claims or platforms that aggregate market signals, the challenge has always been less about generating data than extracting something reliable from it. Over time, the problems became increasingly technical, and I found myself designing systems whose limits I understood in practice but not yet in theory. That's why I'm going back for the foundations.