Physical Intelligence

Build and operate scalable platform infrastructure supporting Physical Intelligence’s AI training, simulation, and robot deployment systems. The role focuses on Kubernetes, cloud platforms, reliability, security, observability, developer experience, and cost-aware infrastructure.

Build the end-to-end API platform that lets external organizations upload data, fine-tune and evaluate Physical Intelligence models, and run low-latency inference for robots. This backend and systems role combines developer-platform ownership, scalable infrastructure, and product engineering for robotics AI.

Build the API platform that makes Physical Intelligence’s robotics foundation models accessible to external developers and organizations. You’ll own data ingestion, fine-tuning, evaluation, low-latency inference, and partner-facing developer tools end to end.

Build full-stack software that lets operators run, monitor, debug, and deploy robots safely across physical environments. This role combines real-time interfaces, on-device and cloud systems, and product ownership for high-stakes human-robot workflows.

Build embedded firmware and Linux-based systems that enable Physical Intelligence’s robots and physical data-collection devices. You’ll work across hardware, software, and research teams on sensor integration, performance, power, and hardware-software debugging.

Build and scale AI-powered internal tools, agents, integrations, and workflows that improve productivity across engineering, research, operations, and recruiting. This role combines hands-on software development with company-wide AI adoption and enablement.

Build and operate the distributed data infrastructure that powers large-scale robot learning, from multimodal data ingestion through training and evaluation. This systems-focused role combines data engineering, storage, and machine learning infrastructure at petabyte scale.

Build and scale the infrastructure powering Physical Intelligence’s large-scale model training, from GPU/TPU orchestration to efficient JAX pipelines. You’ll partner closely with researchers to turn new ideas into reliable, production-grade training runs.