Working on multi-agent safety and cooperative-AI-thingyy. Other interests include s-risks, safe-pareto improvement, and theoretical computer science.