Research Engineer, Infrastructure

Build and operate large-scale distributed training and experiment infrastructure for research.

• •Build and operate large-scale distributed training and experiment infrastructure to accelerate research across thousands of GPUs.
• •Partner closely with researchers to design reliable, high-performance systems for training, rollout, and data pipelines.
• •Key Responsibilities Own distributed training infrastructure: job launchers, checkpointing, fault tolerance, and monitoring.
• •Scale and run massive agent rollouts in VM sandboxes and distributed environments.
• •Profile and optimize training throughput, memory, and communication.
• •Design experiment orchestration, tooling, and high-throughput data pipelines.
• •Diagnose and improve reliability across GPUs, networking, and numerics.
• •Requirements Deep experience building and operating distributed training systems for large models.
• •Strong systems engineering fundamentals across distributed systems, networking, and storage.
• •Proficiency in Python and C++; systems-level experience with PyTorch.