Research Engineer, Infrastructure, Training Systems
Core
Build and optimize distributed training infrastructure for large-scale AI model pre-training and mid-training, ensuring reliability and efficiency across thousands of GPUs.
Role type
Senior IC infrastructure engineer (distributed training systems)
Builds
Distributed training frameworks, runtimes, and workflows for frontier AI models
Domain
AI/ML infrastructure, high-performance computing, distributed systems
Deliverable
production ML models
Required skills
Rust, distributed training frameworks (FSDP, ZeRO), parallelism strategies, GPU communication (NCCL), fault tolerance, deep learning frameworks (PyTorch, JAX), systems debugging
Preferred skills
Low-precision training (FP4/FP8), compiler work, training workload orchestration, open-source contributions
Technologies
Rust, PyTorch, JAX, NCCL, FSDP, ZeRO, Megatron-LM, DeepSpeed
Responsibilities
Design and operate distributed training infrastructure scaling across thousands of GPUs; improve training throughput and accelerator utilization; develop parallelism strategies and runtime components; enhance reliability through observability and checkpointing; collaborate with researchers to validate system changes; build reusable frameworks for experimentation.
Seniority
Senior, hands-on IC
