The U.S. National Science Foundation (NSF) has awarded the National Data Platform (NDP) a five-year, $24.5 million Cooperative Agreement to build a federated, AI-ready national data ecosystem, making it easier to discover, access and analyze data, regardless of where it is stored—and Rafael Coelho Lopes de Sa, associate professor of physics, is playing a key role by leading the development and deployment of the NDP connector nodes.

Image

The project is led by the University of California San Diego in partnership with researchers from the University of Utah, the Texas Advanced Computing Center at The University of Texas at Austin and the University of Washington’s eScience Institute, as well as UMass Amherst and the Massachusetts Green High Performance Computing Center (MGHPCC).

Coelho Lopes de Sa and his colleagues at the MGHPCC will develop and deploy the nodes, which will in turn provide a common layer of computing and software infrastructure to be installed at participating data centers around the country. They will allow NDP to discover and make use of locally operated data, computing and AI services through consistent interfaces, while respecting each institution’s security requirements, operating practices and governance policies. 

Instead of replacing the powerful systems that universities and supercomputing centers have already built, NDP will allow those systems to function as parts of a larger national ecosystem. A researcher could, for example, discover a dataset maintained at one institution, analyze it using an AI or high-performance computing resource at another and share the resulting workflow through a common environment without first having to master the technical details of every participating facility. 

“Research institutions across the country have built extraordinary resources, but researchers should not have to become experts in the internal workings of every facility to use them,” said Coelho Lopes de Sa. “Our goal is to make the boundaries between those facilities largely transparent to the researcher. The connector nodes will provide the common cyberinfrastructure that allows every center to contribute its strengths while retaining control of its own systems.”

“America's leadership in artificial intelligence depends not only on advanced computing resources but also on the data infrastructure that enables researchers to discover, access, share and analyze scientific data at scale,” said Brian Stone, performing the duties of the NSF director, in NSF's announcement of the awards. “These investments provide foundational capabilities that empower scientists and engineers to drive transformative AI-enabled discovery, accelerate innovation and strengthen national competitiveness.”

MGHPCC provides an especially strong foundation for this work because it already brings together an unusually rich and diverse collection of advanced research-computing environments. These include the new MGHPCC AI Computing Resource, a shared high-performance AI environment developed by MGHPCC, its six university members and the Massachusetts AI Hub. UMass Amherst also operates the multi-institutional Unity high-performance computing platform at MGHPCC, while the UMass Physics Department operates NET2, a major computing and data facility that supports the ATLAS experiment at CERN, both of which are used for AI development and deployment. By making such resources more discoverable, interoperable and reusable, NDP can help researchers spend less time navigating infrastructure and more time turning data into scientific knowledge.

“By uniting the expertise of SDSC, the University of Utah, the University of Massachusetts Amherst, TACC and UW with NSF's broader investments in cyberinfrastructure and the National Artificial Intelligence Research Resource, we are excited to lead NDP's efforts to ensure that the benefits of AI-driven discovery are accessible to the full breadth of the U.S. research community,” said the University of San Diego’s Frank Wuerthwein, who directs the university’s Computing San Diego Supercomputer Center.