Speaker
Description
The DOE Nuclear Physics AI-Ready Accelerator Data (NARAD) project is building a facility-agnostic way to describe accelerator facilities so that machine-learning tools, agentic assistants, and digital twins can find and use their data. NARAD extends the Particle Accelerator Language Standard (PALS): a flattened PALS lattice supplies the physics model, and NARAD adds facility configuration data (including control system signals, magnet transfer functions, units, and provenance) in a LinkML schema that validates instances and produces a knowledge graph. We report progress in the project's second year. NARAD instances now describe the CEBAF injector, the UITF injector, and the Brookhaven Booster-to-AGS (BtA) transfer line, with BtA quadrupole and corrector scan data are published on Zenodo with an American Science Cloud data card. The UITF instance supports channel-finding queries for an LLM-based controls assistant, and benchmark questions are being used to measure query accuracy. We describe lessons learned on units, naming, and schema validation, and discuss the remaining steps toward the EIC: ring and recirculation topology, scaling to large lattices, and linking EIC accelerator models to operational data.