Research Engineer, Atlas Physics Simulation for RL at Boston Dynamics

Candidates should have an MS with 3 years of industry experience or a PhD in Computer Science, Machine Learning, Robotics, or a related field. A detailed understanding of physics simulation, including contact solvers and mesh representations, is required. Extensive experience with physics simulation tools such as MuJoCo, IsaacSim, and Warp, along with experience in rendering pipelines for photorealistic synthetic images, is necessary. A strong foundation in Python, C++, and modern numerical frameworks like PyTorch and Jax, as well as experience in algorithm design, debugging, and performance optimization, are essential. An ideal candidate would have a PhD or equivalent research experience in reinforcement learning or robotic manipulation, publications at top robotics venues, experience training RL policies in simulation, a strong understanding of GPU-based simulators, and knowledge of GPU/CPU compute architectures and heterogeneous compute clusters, Kubernetes, and Docker.