Research Engineer

Make our models performant, fast, and reliable at scale by developing the core frameworks used to train and evaluate our ML models.

• Build datasets and algorithms to systematically evaluate and monitor model performance.
• Analyze model failure modes, working closely with researchers on experiments to mitigate them.
• Own the distributed training stack, removing runtime and reliability bottlenecks across all layers of the stack and architecture—parallelism, quantization, kernel optimization and whatever else might move the needle.
• Own and contribute to our auto research and auto kernels frameworks.