AutoRobots

An autonomous AI scientist that teaches simulated robots to move. No human reward engineering. The LLM agent iterates on reward functions, trains RL policies, evaluates results, and accumulates improvements — all on Nebius infrastructure.