Software Engineer, Trainium

Responsibilities

• Build and optimize OpenAI’s inference stack for AWS Trainium.
• Develop high-performance kernels for critical model operations and workloads.
• Extend compiler support to efficiently target Trainium hardware.
• Build systems to execute and optimize model forward passes on Trainium.
• Profile workloads and identify bottlenecks across kernels, compiler-generated code, runtimes, and model execution.
• Partner with inference and ML systems teams to bring new models and architectures to Trainium.
• Solve hardware/software performance problems end-to-end from investigation through production deployment.

Requirements

• 3+ years of relevant engineering experience, ideally in ML systems, compilers, kernels, runtimes, or performance engineering.
• Strong systems programming fundamentals and experience writing performance-critical software.
• Experience with GPU, TPU, Trainium, or other specialized accelerator architectures.
• Ability to reason about performance across hardware, kernels, compilers, and ML frameworks.
• Ability to own technically ambiguous problems end-to-end and learn new hardware and software domains.
• Experience with AWS Trainium or the AWS Neuron SDK is a bonus.
• Contributions to PyTorch or JAX, compiler infrastructure such as LLVM or MLIR, XLA, Triton, or specialized accelerator kernel development are bonuses.