Training Performance Engineer
San Francisco
Workplace: HybridFull timeUSD 250,000 - 445,000 annuallyFunction: Education & TrainingSkills: ["Python","C++","Rust","CUDA","PyTorch","JAX","TensorFlow","GPU","Multi-GPU","HPC","Profiling","Throughput","Distributed training","Model training","Kernel efficiency","Scheduling","Communication","NCCL","MPI","UCX"]Join OpenAI as a Training Performance Engineer to optimize large-scale distributed model training. You’ll profile training runs, boost GPU utilization, reduce bottlenecks across compute, communication and storage, and collaborate with runtime and systems teams to push throughput and uptime for frontier-scale models in a hybrid San Francisco, CA setting.
Loading
Loading job details...
Preparing the role view and application actions.

