高级开发工程师，物理AI基础设施

The Infra-Compute division builds large-scale, highly available cloud and AI infrastructure that powers our public cloud offerings and internal products.

Our US team develops technologies across AI training, inference, and agent infrastructure.

We are expanding this work into Physical AI: intelligent systems that perceive, reason, and act in the physical world.

The team focuses on infrastructure for Physical AI, including data and simulation platforms, distributed training and inference, and deployment on heterogeneous hardware.

We collaborate closely with customers, researchers, open-source communities, and hardware partners to turn emerging research into reliable, scalable systems.

- Independently own the design, implementation, validation, and operation of major components within a Physical AI platform, focusing on one or more areas across data pipelines, simulation, model training, evaluation, inference, and heterogeneous deployment.

- Develop scalable infrastructure for training and serving multimodal and robotics models, including vision-language-action models, world models, and robot policies.

- Translate research ideas into production-ready capabilities, with clear reliability, performance, and quality targets.

- Lead technical design for complex projects, make well-reasoned trade-offs across performance, reliability, cost, and delivery speed, and drive projects through production adoption.

- Collaborate with adjacent teams, mentor engineers, and contribute to shared architecture and engineering standards.

- Bachelor's degree or above in Computer Science, Computer Engineering, Robotics, Electrical Engineering, Artificial Intelligence, or a related technical field, or equivalent practical experience.

- 5 years of software engineering, machine learning systems, robotics, or related industry experience.

- Strong software and systems engineering fundamentals, with experience designing, building, and operating production-scale distributed or data-intensive systems.

- Robotic learning, including imitation learning, reinforcement learning, vision-language-action models, world models, or policy evaluation.

- AI training and inference infrastructure, including distributed training, inference engines, GPU kernels, or collective communication.

- Robotics data and simulation systems, including multimodal dataset verification, synthetic data generation, simulation, data curation, or sim-to-real workflows.

- Contributions to relevant open-source projects such as Isaac Lab, Isaac Sim, MuJoCo, Cosmos, vLLM-Omni, SGLang Omni, RLInf, etc.

- Experience building large-scale AI or robotics platforms used by multiple teams or external customers, or experience with real-world robot systems and operational challenges.