As the Director of the Applied Mathematics and Computational Research Division at Lawrence Berkeley National Laboratory (Berkeley Lab), Stefan Wild leads teams of computational scientists, computer scientists, and mathematicians in the research and development of mathematical models, high-performance software and algorithms, computer system architecture, and data-driven technologies to solve some of the world's most challenging computational problems across the disciplines, including materials science, biology, and physics.
In addition to his leadership role at Berkeley Lab, Wild is also contributing his expertise to the Genesis Mission, a national initiative to advance artificial intelligence in order to accelerate scientific discovery and deliver solutions to challenges in science, energy, and national security. A cornerstone of the Genesis Mission's AI models and data efforts is the Transformational AI Models Consortium, or ModCon. ModCon will build and deploy self-improving AI models by harnessing DOE's unique data, facilities, and expertise. Wild leads one of ModCon's four core teams: Baseline AI R&D Capabilities, or BASE. This team develops the reusable methods and structure that form the foundation of every Genesis Mission science project.
In this Q&A, Wild shares his views on ModCon's essential role in accelerating scientific discovery and innovation, and how Berkeley Lab's unique capabilities are critical to advancing it.
Q: Why is ModCon integral to the Genesis Mission?
Wild: ModCon provides a central set of forward-leaning AI capabilities across the Genesis Mission. In close collaboration with the American Science Cloud and the private sector, we're enabling the national science & technology challenge teams to accelerate domain breakthroughs and demonstrate AI advantage. Key components of ModCon are a "BASE" team, which serves as Genesis Mission's AI R&D team; a Best Practices for Scientific Workflows (BPSW) team for integrating AI into scientific workflows; a Data Brokering and Standards (DBS) team working on standards and governance for centralized access to Findable, Accessible, Interoperable, and Reusable datasets; and an Intellectual Property and Partnerships Formation (IPPF) team to streamline lab-industry-academia collaboration.
"Over the past decade, Berkeley Lab researchers have been making advances in scientific machine learning, from mathematical foundations to scalable software and breakthroughs across the sciences." - Stefan Wild, Applied Mathematics and Computational Research Division Director and ModCon BASE team lead
Q: How is Berkeley Lab playing a significant role in the ModCon effort?
Wild: Berkeley Lab is leading ModCon's BASE effort, which is rapidly standing up core AI capabilities that cut across many Genesis Mission teams. Within BASE, Berkeley Lab's Shreyas Cholia leads the BASE/Data thrust, which is developing the Data Science Agent Toolkit, an AI-agent-driven data curation pipeline framework that assesses AI-readiness and provides tools to autonomously transform raw scientific data into training-ready form. I also lead the BASE/Eval thrust, which builds the Genesis Mission Proving Ground, a comprehensive suite for evaluating and tracking the progress of AI models and agentic workflows along many customizable metrics, from those capturing robustness to those measuring multiple modes of scientific validity.
To the DBS effort, we are also contributing data detectives - researchers who identify and characterize data sources, with Berkeley Lab's team focused on biological, materials, and applied energy data - as well as standards for AI readiness. Meanwhile, the Lab's work in BPSW is capturing much of the rapidly evolving best practices in increasingly agentic-forward workflows. Finally, the IPPF effort draws on the Lab's expertise in research, facilities, and sponsored projects to accelerate collaboration with a broad range of partners.
It is a real team effort: every science area at Berkeley Lab has been involved in ModCon activities.
Q: How does Berkeley Lab's expertise and capabilities make the lab well-positioned to contribute to the Genesis Mission?
Wild: Over the past decade, Berkeley Lab researchers have been making advances in scientific machine learning, from mathematical foundations to scalable software and breakthroughs across the sciences. Berkeley Lab has also led data management best practices in many fields and delivered tools that address national priorities, such as through DOE's Exascale Computing Project. Coupling this expertise with our excellent user facilities and the ScienceIT team's efforts to bring frontier models to every employee, Berkeley Lab is in an exceptional position to deliver ambitious scientific breakthroughs as envisioned by the Genesis Mission.
