Building AI-Ready Synchrotron Data Ecosystems with NOMAD

This talk was given on September 03, 2026 at the HZB-KFS BESSY III User Workshop (https://events.hifis.net/event/3728). In this contribution, we discuss approaches developed within FAIRMat and the NOMAD ecosystem for building FAIR data infrastructures for future synchrotron environments. The presentation outlined a layered approach spanning structured and harmonized data, scalable infrastructure, provenance and reproducibility, and model and AI operation. Particular emphasis was placed on standardized metadata, semantic data models, and community standards such as NeXus as foundations for interoperable and machine-actionable experimental data. Building on these foundations, examples from FAIRMat and NOMAD illustrated how data infrastructures can connect acquisition systems, analysis workflows, computational environments, and long-term repositories while maintaining provenance and reproducibility. The contribution further discussed the implications for AI-supported and increasingly autonomous experimentation. As future synchrotron facilities move toward higher throughput, multimodal measurements, and closed experimental loops, FAIR data infrastructures need to support not only researchers but also automated workflows and AI systems. The presentation therefore highlighted the opportunity to establish interoperability, reproducibility, and machine-actionable data as integral components of the design of future facilities such as BESSY III. As part of the German National Research Data Infrastructure (NFDI), FAIRMat contributes to a broader national initiative for sustainable and interoperable research data management across scientific communities. This work is funded by the NFDI consortium FAIRmat - Deutsche Forschungsgemeinschaft (DFG) - Project 460197019.