PyTorch is looking for strong Postdoctoral Researcher candidates to work on optimization techniques for AI Accelerators. Our team works on Accelerator Performance for Deep Learning workloads across the stack - performance tooling, benchmarking, kernel authoring, compilation, and enablement. Our work is key to Meta making efficient use of its AI fleet, and ML Engineers being able to move fast with model exploration, without the friction of performance considerations. We have strong partnerships with customer teams in GenAI, recommendation systems, and content understanding. A Postdoctoral Researcher in PyTorch typically works in a manner similar to a researcher at a university. We provide an environment where the researcher can draw inspiration from the problems Meta faces - scale, complexity, and constant change. The specific research area may include 1) the application of compiler techniques targeting AI Accelerators and 2) optimization techniques for training and inference of very large models. For this position, the researcher will engage with the supervisor and other researchers and engineers to create and implement a research agenda and publish world-class research.