DOE Genesis Mission Awards UC San Diego $M for AI Physics Tools

The University of California San Diego will lead development of an AI system designed to autonomously analyze data, design models, and deploy them on specialized hardware for physics and quantum computing research, according to details first reported by UC San Diego Today.

The project, called SIDERIUS, is one of 278 awards announced under the Department of Energy's Genesis Mission initiative. Assistant Professor Aobo Li from UC San Diego's Department of Physics and Halıcıoğlu Data Science Institute serves as principal investigator, with Associate Professor Yi-Zhuang You as co-PI.

What SIDERIUS will do

The system aims to function as an AI scientist capable of translating scientific challenges into formats that machine learning systems can tackle effectively. Rather than requiring researchers to manually prepare data and configure models, SIDERIUS will use large language model agents to handle those steps automatically.

"The next frontier of AI for science is learning how to translate scientific challenges into a form that AI can solve effectively," Li said. The team's long-term vision extends beyond physics to any data-intensive scientific domain.

Four additional collaborative projects

Beyond leading SIDERIUS, UC San Diego researchers will contribute to four Genesis Mission projects led by other institutions:

• CERN particle collision analysis: Associate Professor Javier Duarte and computational specialist Daniel Diaz will work with the University of Colorado Boulder on ultra-fast AI tools to process massive data streams from the Large Hadron Collider.

• Neutrinoless double beta decay search: Li will collaborate with UNC Chapel Hill to deploy AI agents that automate optical simulations for the LEGEND experiment, which searches for an extremely rare nuclear process that could explain why the universe contains more matter than antimatter.

• Unified foundation model: Professor Liang Yang and Li will work with the University of Hawaii to build a foundation model trained on datasets from five different neutrinoless double beta decay experiments.

• Real-time signal processing: Li will partner with Pacific Northwest National Laboratory on AI tools that identify genuine decay signals while filtering background noise from the LEGEND detector in real time.

Why it matters

The Genesis Mission represents the DOE's largest-ever research funding competition, with more than 5,000 submissions and a 6% acceptance rate. UC San Diego's involvement in five projects—one as lead institution—demonstrates how AI infrastructure and interdisciplinary collaboration are becoming critical differentiators in scientific computing. The SIDERIUS approach of automating the translation between scientific problems and AI-ready tasks could accelerate discovery timelines across multiple research domains, from particle physics to materials science, by reducing the manual overhead currently required to apply machine learning to complex datasets.

The Genesis Mission aims to combine DOE scientific capabilities, advanced AI, and high-performance computing to transform scientific discovery and strengthen American technology leadership. The 278 awards will address challenges including nuclear energy, critical mineral extraction, intelligent chip design, and commercial fusion energy.

These details were first reported by UC San Diego Today.