People

Omkar Salpekar

Software Engineer

I am a Software Engineer on the PyTorch Distributed team in Facebook AI Infrastructure, where I focus on large-scale distributed training for a variety of models from ranking to CV and NLP. Specifically, I am building data-, model-, and hybrid-parallel techniques in PyTorch to support training complex models across both commodity and high-performance computing clusters.

Prior to Facebook, I studied Electrical Engineering and Computer Science at UC Berkeley, where I worked on a distributed data analytics library based on Ray at the RISELab.

Interests

Machine learning, distributed systems

Related Links

Google Scholar

Latest Publications

PyTorch Distributed: Experiences on Accelerating Data Parallel Training

Shen Li, Yanli Zhao, Rohan Verma, Omkar Salpekar, Pieter Noordhuis, Teng Li, Adam Paszke, Jeff Smith, Brian Vaughan, Pritam Damania, Soumith Chintala

VLDB - August 31, 2020