Data Management for AI-Powered Science Agents: From Storage Optimization to Context

Abstract: The advent of agentic artificial intelligence (AI) promises to revolutionize scientific discovery, but this potential is often crippled by a fundamental mismatch: our data infrastructure, designed for human-driven, bulk-synchronous workflows, is ill-equipped to serve autonomous agents that require dynamic, context-aware information.

This talk explores the critical transition from passive storage optimization to active context orchestration. I will introduce IOWarp (iowarp​.ai), an NSF-funded data management platform developed at Illinois Tech designed to bridge this gap. IOWarp treats data, metadata and computational state as a unified, intelligent context, actively assembling and delivering the right information from multi-modal sources — simulations, sensors and literature — to AI agents.

By transforming our approach to data management, we can unlock truly autonomous scientific workflows, turning AI agents from expensive ​“data janitors” into powerful research partners and maximizing the scientific return from Argonne’s world-class facilities.

Bio: Anthony Kougkas is an associate research professor at the Illinois Institute of Technology and deputy director of the Gnosis Research Center. He also serves as a guest research faculty member at Argonne.

See upcoming and previous presentations at CS Seminar Series.