Research Member of Technical Staff- Training Systems

Highlights

• Own training performance end-to-end for generalist intelligent robots.
• Drive scalability of multimodal training across thousands of GPUs.
• Accelerate model iteration for real-world embodied intelligence.

What they offer

• Equity

Ideal candidate

Systems-minded performance owner who thrives debugging large-scale training and turning bottlenecks into measurable speedups. You love partnering with researchers and owning efficiency end-to-end in a fast-moving team.

Requirements

• Proven track record improving large-scale distributed training performance
• Deep hands-on experience with modern ML stacks (PyTorch required; JAX a plus)
• Strong understanding of data / tensor / pipeline parallelism and sharded training (FSDP / ZeRO-style)
• Strong systems intuition across compute, communication, and memory bottlenecks
• Exceptional debugging and measurement ability turning slow training into validated improvements
• High ownership mindset and comfort in a fast-moving environment