Jeff Larkin is the Director of Supercomputing Solutions Architecture at NVIDIA, where he leads a team of SAs responsible to aiding customers in integrating NVIDIA technologies. He is passionate about the advancement and adoption of parallel programming models for High Performance Computing. He has a long history in performance analysis and optimization of high performance computing applications as a past member of NVIDIA’s Compute Developer Technology group and Cray’s Supercomputing Center of Excellence at Oak Ridge National Laboratory. Jeff is also the chair of the OpenACC technical committee and has worked in both the OpenACC and OpenMP standards bodies, among other open source and open standards engagements. Jeff holds a B.S. in Computer Science from Furman University and a M.S. in Computer Science from the University of Tennessee, where he was a member of the Innovative Computing Lab.
Presentation Title:
Accelerating Science & Engineering with Agentic AI
Presentation Abstract:
Recent advances in AI models, tooling, and agentic workflow design are making AI increasingly useful as a practical accelerator for science and engineering, with clear relevance to ambitious efforts such as the U.S. DOE’s Genesis Mission. This talk will survey how modern AI systems can support scientific software development, simulation workflows, physics surrogate models, physics-informed and physics-guided machine learning, literature synthesis, and autonomous discovery pipelines. Two examples will ground the discussion: a proof-of-concept scalable, GPU-resident hydrodynamics code co-developed with Codex, and an agentic discovery workflow inspired by LANL work on identifying new Targeted Alpha Therapy drug candidates. Rather than replacing scientists and engineers, these systems can expand what expert practitioners are able to accomplish by combining models, tools, simulations, data, and human judgment in workflows designed for verification, reproducibility, and scientific rigor.