Speaker
Description
Real-time AI/ML is emerging as a critical capability for future high-rate nuclear physics experiments, such as EIC, where rare physics signals must be identified within complex detector data streams. This talk will summarize key lessons learned from the DOE-funded project “Intelligent Experiments through Real-Time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors,” focusing on the sPHENIX fast-ML heavy-flavor trigger demonstrator development using graph neural networks on FPGA. The project provided important experience in physics-driven model design, latency and resource constraints, model compression, hardware-aware optimization, and AI/FPGA co-design for streaming readout. Building on these lessons, I will discuss the new Genesis Phase-I project, “Agentic AI for Real-Time Expedited Discovery from High-Complexity EIC Data Streams,” which offers an opportunity to develop agentic-AI workflows for EIC data processing using simulated ePIC DIS events and beam-induced backgrounds, with goals of physics-preserving background rejection, reconstruction-aware validation, and accelerated optimization toward future real-time intelligent experiments.