17–22 May 2026
C.I.D
Europe/Zurich timezone

Project Poseidon: Foundation Models and Agentic Workflows through Unified Control System Data Architecture

MOP6323
18 May 2026, 16:00
2h
C.I.D

C.I.D

Deauville, France
Poster Presentation MC6.D13: Instrumentation: Artificial Intelligence Poster session

Speaker

Thorsten Hellert (Lawrence Berkeley National Laboratory)

Description

Modern accelerators face data challenges exceeding manual analysis. Fragmented architectures (e.g., EPICS) further hinder ML model generalization. We present Project Poseidon, a unified platform designed to standardize large time-series analysis for Foundation Models and Autonomous Agents.
Poseidon's control-system-agnostic framework includes a unified database, modular analysis engine, and facility adoption layers. This decoupling enables multi-facility data aggregation, essential for training Large Observation Models (LOMs). We discuss Poseidon's role in enabling agentic workflows to monitor large systems exceeding 50,000+ variables, detect complex correlations, and autonomously trigger corrections, shifting operations to intelligent, predictive oversight.

In which format do you inted to submit your paper? LaTeX

Author

Thorsten Hellert (Lawrence Berkeley National Laboratory)

Co-authors

Antonin Sulc (Lawrence Berkeley National Laboratory) Gianluca Martino (Lawrence Berkeley National Laboratory) Simon Leemann (Lawrence Berkeley National Laboratory)

Presentation materials

There are no materials yet.