Argonne Leadership Computing Facility

This summer, Abrar Faiyad, a Ph.D. student in computational materials science at UC Merced, is working at Argonne National Laboratory to develop autonomous AI workflows for training machine-learning interatomic potentials on the ALCF’s Aurora supercomputer. By combining foundation models, AI agents, exascale computing, and experimental validation, the workflows aim to enable more efficient simulations of complex reactive materials, helping accelerate next-generation reactor design, electrocatalyst discovery for hydrogen production, and broader advances in materials science.

“The two things that have changed me most this summer are scale and automation — and how the two together completely redefine what a single researcher can attempt,” Faiyad said. “Watching a pipeline I built launch DFT calculations, train a model, validate it, identify its own blind spots, and then go back for more data — without me touching it — feels like a glimpse of what science is going to look like in the next decade. It is the moment where leadership computing stops being a faster calculator and starts being a genuine collaborator.”

Faiyad is part of a talented group of working on research projects at the intersection of high-performance computing, AI, and science. Follow along this summer as we highlight their projects and experiences.

Learn more about student opportunities at ALCF: https://lnkd.in/ggN2E4Vv

This Fourth of July, we’re proud to celebrate America’s 250th anniversary alongside two milestones in our own history: Argonne National Laboratory’s 80th anniversary and the ALCF’s 20th year of advancing science through high-performance computing.

Happy Independence Day from the ALCF!