A funding surge for AI-driven discovery
George Mason University computer science professor Keren Zhou has been selected by the U.S. Department of Energy for a Genesis Award, part of the Trump Administration’s national Genesis Mission. The initiative aims to build an integrated scientific discovery platform that unites national laboratories, universities, industry partners, and philanthropies to drive breakthroughs in energy, discovery science, and national security.
Nearly 300 projects were selected nationwide after what the DOE described as the largest response to a funding opportunity in its history. Zhou’s project received $750,000 in Phase I funding for a nine-month research period.
Zhou’s framework: from months to seconds
His project targets a bottleneck that affects nearly every field that relies on supercomputers, from climate science to drug development. Today, even small changes to scientific software can require days or months of trial-and-error testing to make sure programs run efficiently on different machines. That slows down research and delays discoveries.
“Today’s HPC applications may need days or even months of tuning every time a parameter changes or a new hardware configuration is introduced,” Zhou said. “We want to eliminate that costly cycle by building a proxy model that can rapidly estimate performance—reducing simulation time from months to seconds.”
In practical terms, the system acts like a speed-check for scientific computing. Instead of running full, time-consuming tests, researchers could quickly predict how their programs will perform and automatically identify faster, more efficient ways to run them. With AI systems now capable of rewriting code and suggesting improvements, Zhou’s framework could help scientists optimize their work in minutes rather than weeks.
National lab partnerships and student access
Zhou is the sole George Mason investigator on the project but will hire postdoctoral researchers as the work advances. Through the Genesis Mission’s shared platform, students will gain access to industrial partners such as AWS, OpenAI, and Google, along with AI tokens and cloud computing resources. He is also collaborating with Oak Ridge National Laboratory and Brookhaven National Laboratory.
As Phase I progresses, Zhou aims to show that his approach works across different scientific applications and hardware systems and delivers significantly faster performance without requiring long simulation runs. Strong results would position his team for Phase II funding, which the DOE expects to award next year.
This type of work sits at the center of a growing effort to apply machine learning directly to the process of science itself, an area often described as AI for Science & Research. “The Genesis Mission rests on a clear premise: AI is becoming an instrument of science, not only an object of it,” said Amarda Shehu, George Mason’s vice president and chief AI officer. “Professor Zhou’s framework is a clear case, a model that predicts and optimizes the systems on which discovery now depends, compressing months of tuning into seconds.”
Andre Marshall, George Mason’s vice president for research, innovation, and economic impact, said the award underscores the university’s growing role in advancing the nation’s scientific and technological capabilities. “Professor Zhou’s project elevates our visibility in the federal research ecosystem and demonstrates how George Mason is contributing to the next generation of AI-driven discovery.”
Why this matters for science and research professionals
For researchers who spend weeks tuning code for different supercomputing environments, Zhou’s proxy model promises to cut that cycle to seconds. If the framework generalizes across applications and hardware, it could eliminate a major hidden cost in computational science: the time spent manually optimizing software for each new machine or parameter change. Phase II funding and broader adoption would depend on results due next year.
Your membership also unlocks:
