Aignosis: Empowering Early Detection

I grew up inside my father's medical practice in Jaipur, close enough to watch it from the inside. What stayed with me wasn't the illnesses themselves but the timing: parents describing a child who wasn't quite meeting a milestone, less eye contact than expected, a word that hadn't come yet, and being told, gently, to wait and watch. By the time anyone acted, the child was often four or five, well past the window when early support tends to matter most.

I watched that pattern repeat, consultation after consultation, until it stopped being something I noticed and became something I couldn't stop thinking about.

I had an offer from Kellogg's. A stable, well-worn path. I turned it down anyway, because I'd already seen what the cost of waiting looks like on a family.

I'm a physicist who fell for artificial intelligence. I was standing at the entrance to a PhD, the kind of opportunity most researchers spend years working toward, when I chose to walk through a different door instead.

I staked my training on a harder, messier question: could a laptop webcam notice, in a few minutes, what an overstretched healthcare system was taking years to catch?

I refused to take shortcuts. Our models don't just glance at a child's gaze, they read it as a signal unfolding moment by moment. I tested that against simpler methods, again and again, until I trusted it actually worked.

For me, it wasn't about testing a model. It was about trying to give a child back the years “wait and watch” was costing them.