About

I’m a physicist turned AI researcher, turned product manager. What’s next? Let’s see.

I believe that physics gives you the best mental models to solve problems, and I try to apply this thinking to many other fields.

My current interests are how to make autonomous agents work in unpredictable environments, continual learning, and human modeling.

The latter asks a fundamental question: how can we predict what a human will do, and what is the simplest model that describes it? (And by “simple,” I mean it might still take a trillion-parameter neural net!). Today’s AI systems are inflated with bias and by no means model true human behavior. Solving that will be incredibly valuable.

This blog is a collection of my recent thoughts. Feel free to reach out!