SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
OpenAI's Inference team is seeking a Software Engineer to lead optimization of frontier AI models on TPUs. You will own performance across the complete forward pass, from model code through kernels to hardware execution.
In this role, you will optimize model execution on TPUs, improving performance and hardware utilization across the full stack. You'll profile workloads to identify bottlenecks, implement improvements to latency and throughput, and develop optimized kernels guided by understanding of surrounding model architecture. You'll explore efficiency techniques such as quantization while preserving model quality, build benchmarks and tools to measure performance and validate correctness, and partner with researchers and engineers to support new model architectures and bring improvements into production.
You'll have broad technical ownership: identifying performance bottlenecks, choosing the most impactful improvements, and carrying them through implementation and validation. This is a senior individual contributor role with significant technical scope and influence across the inference stack.
REQUIREMENTS:
- Practical experience optimizing machine learning workloads on TPUs (training, inference, or both)
- Understanding of modern model architectures and how computation maps onto accelerator hardware
- Ability to connect detailed kernel behavior to complete model performance and work across stack layers
- Strong software engineering skills with systematic approach to profiling, debugging, and validating performance improvements
- Ownership of open-ended technical problems with ability to identify priorities, make design decisions, and deliver solutions
- Collaboration skills with researchers and engineers; comfort learning unfamiliar system parts and communicating technical tradeoffs
- Experience with quantization or attention kernels (especially valuable)
