Piyush K. Sao

Hello! I’m Piyush Sao, a staff scientist at Oak Ridge National Laboratory (ORNL) in the Computer Science and Mathematics Division. My research focuses on improving scientific computing and machine learning algorithms by leveraging high-performance computing platforms like the Frontier Supercomputer at ORNL.

I develop efficient numerical algorithms, including linear solvers and graph algorithms for traversal, shortest path, and clustering. Some of my algorithms have been implemented in widely-used libraries such as SuperLU_DIST and ArborX. My work addresses fundamental questions that push the boundaries of modern computing:

1. How can we create algorithms that efficiently scale on the world’s most powerful supercomputers?
2. What is the minimum amount of resources required to accurately and reliably solve complex problems?

By tackling these challenges, I strive to develop communication-avoiding algorithms, improve fault tolerance, and optimize scientific applications for parallel and distributed computing environments.

I completed my undergraduate studies at IIT Madras in India and earned my PhD from Georgia Tech. At Georgia Tech, I was part of HPC Garage, a research group led by Prof. Richard Vuduc.