My interests are broadly in theoretical computer science, especially questions at the interface of learning, probability, and computation. More recently, my research has centered on the foundations of trustworthy AI: understanding when learning systems can make reliable predictions and how that reliability can be quantified.
More generally, I like to think in terms of technique. I am especially interested in questions that bring together tools from diverse areas of mathematics, and I am always happy to think about problems where a theoretical lens may be useful.
Earlier, I worked on problems in complexity theory, often using algebraic techniques. I am very interested in how methods from different areas can interact to shed light on fundamental questions.