About this role
Physical Superintelligence is hiring AI researchers to build and train AI agents and training systems that learn to do physics, design evaluation workflows and benchmarks, and make agent behavior interpretable while collaborating with physicists and engineers. The role is based in Boston (remote candidates considered case-by-case) and requires a PhD-level research track record, hands-on RL experience, and proficiency with modern ML frameworks and distributed training.
Skills for this role
Reinforcement LearningAI agentsAgent-based systemsPPOSACMuZeroMulti-agent self-playSearch-augmented methodsPyTorchJAXDistributed trainingRayXLAAccelerateScaling ML to clustersExperimental methodologyBenchmark designEvaluation workflowsInterpretabilitySimulatorsScientific tool integrationPhysics domain knowledgeMathematical modelingPublishing in top ML venues