Nicholas A. Nystrom is Chief Scientist at the Pittsburgh Supercomputing Center (PSC), which is a joint research center of Carnegie Mellon University (CMU) and the University of Pittsburgh, and Visiting Research Physicist at CMU. He is architect and principal investigator (PI) for PSC's Bridges ($21.9M award to date) supercomputer, which pioneered the convergence of HPC, AI, and Big Data, as well as for Bridges-2 ($10M acquisition), which introduces new innovations and much greater capacity to support rapidly evolving research, particularly research involving HPC, AI, and Data. Bridges and Bridges-2 are supported by the U.S. National Science Foundation (NSF) and are available at no cost for research and education. Nicholas is also PI for the Human BioMolecular Atlas Program (HuBMAP) central Infrastructure and Engagement Component, which is developing a map of tissues in the human body at single-cell resolution. HuBMAP is a collaboration currently involving over 50 institutions worldwide and supported by the U.S. National Institutes of Health (NIH). He is co-PI for its upcoming Neocortex AI system and co-PI for Open Compass, which is exploring the potential of emerging AI technologies for research applications. Nicholas also leads projects to enable sharing and reproducing computer architecture simulations and for sharing data and tools relating to coastal water modeling.
Nicholas's current research interests focus on computer architecture for high performance AI and HPC, applications of AI to medical image and genomic data, and enabling data discovery and interoperability. His academic background includes a Bachelor’s degree in Chemistry, Math, and Physics and a Ph.D. in Quantum Chemistry from the University of Pittsburgh. When not working, he enjoys rock climbing and time in mountains and wilderness.
Topic: Accelerating AI for Data-Driven Scientific Discovery
• Nicholas A. Nystrom, Chief Scientist, Pittsburgh Supercomputing Center & Carnegie Mellon University
• Paola A. Buitrago, Director, AI & Big Data, Pittsburgh Supercomputing Center & Carnegie Mellon University
Artificial Intelligence (AI) is enabling great breakthroughs across research, yet its computational needs are outpacing mainstream computing. How can we continue to drive the breakthroughs we need to combat pandemics, develop effective treatments for cancer and rare diseases, protect the environment, and create a sustainable future? Our approach at the Pittsburgh Supercomputing Center brings together revolutionary computer architecture that rejects previously assumed constraints to achieve unprecedented levels of performance for specific domains, coupled with general-purpose high-performance computing (HPC) and high-performance data analytics (HPDA). We describe Neocortex, a highly innovative system designed to introduce the Cerebras CS-1 and Wafer Scale Engine – the largest processor ever built with 1.2 trillion transistors and 400,000 cores – to the research community, and moreover, to enable scaling ambitious deep learning training across multiple CS-1 systems by leveraging a 24 TB HPE Superdome Flex as an unusually capable “front end”. We also describe Bridges-2, another new supercomputer with which Neocortex will be closely federated to provide large-scale data management, preprocessing, and complementary simulation capabilities. Both systems will be available at no cost for open research.