About

I lead the Human Aspects of Machine Learning group at the Max Planck Institute for Intelligent Systems in Tübingen. In my work these days, I study the following question: when we say a model "understands," "reasons," or "scores well," is the test behind that claim actually valid and reliable? Does it measure what it claims to, and does it give consistent results?

I'm an ML scientist by training: I hold a BSc in pure mathematics and an MSc and PhD in machine learning. My background has given me expertise across a range of problems, including fair, trustworthy, and explainable ML; building effective, efficient hybrid human-ML collaborative systems; and, these days, benchmarking and evaluating AI.