Assistant professor of computer science at ETH Zurich. His current work studies the worst-case behavior of Deep Learning systems from an adversarial perspective, to understand and mitigate long-term threats to the safety and privacy of users.