Software Engineer, TPU Compilers

What you’d do

Optimize OpenAI's frontier models on TPUs, owning forward-pass performance from model code through kernels to hardware. Profile workloads, find bottlenecks, and implement latency and throughput improvements. Found on the employer's careers page on Oct 20, 2026. Apply through the link.

You’ll need

• Strong foundation optimizing real workloads on TPUs
• Kernel development and optimization experience
• Performance profiling and bottleneck analysis
• Experience with training or inference workloads