Presenter

Avinash is a Ph.D. student at Rochester Institute of Technology (RIT), New York. He is broadly interested in large-scale data management for high-performance parallel and distributed computing. Specifically, his work focuses on checkpointing and restoring GPU-resident distributed data structures under concurrency at scale. In addition to fault tolerance, his work facilitates key I/O challenges associated with reproducibility, steering experiments based on analytics of intermediate results, adjoint computations that revisit previous states to advance computations, etc.