About me
Currently, Sanket serves as a Research Engineer at Meta's SuperIntelligence Lab, in PyTorch Distributed and Compiler team. He specializes in performance optimization of large scale training of LLMs based on Mixture of Experts architectures.
Prior to this he obtained his PhD in AI from Harvard University, in 2025, with a focus on creating frameworks for the automated and efficient scaling of distributed deep learning training He has co-authored TorchTian and AutoParallel projects at PyTorch.