Working on multi-agent safety, hoping for the least-bad future. Other interests include s-risks (safe-pareto improvement) and theoretical computer science.