Research Engineer

What you'll do

• Develop distributed training systems across hundreds of GPUs, including parallelization, sharding, and compute optimization.
• Build workflows and tools for data ingestion, experiment tracking, monitoring, debugging, and training performance.
• Resolve training-stack bottlenecks and provide lightweight production inference and cloud infrastructure support.

What they ask for

• Strong experience building infrastructure for large-scale ML training.
• Deep understanding of training and scaling modern LLM/VLM systems.
• Proven experience setting up and scaling distributed training across hundreds of GPUs, with knowledge of data, model, and pipeline parallelism.
• Strong Python proficiency and expert-level PyTorch and/or JAX proficiency; understanding of attention optimization, kernel fusion, and efficient memory usage.

Nice to have

• Experience supporting production inference systems or building experiment-management and researcher-productivity tools.
• Familiarity with robotics or embodied AI workloads.

Tools

Python, PyTorch, JAX

AI focus

Build infrastructure for training and deploying generalized physical AI models for industrial robotics, including large-scale LLM/VLM training and inference optimization.