AI RCS Strategy Workshop

from Monday, 15 September 2025 (08:09) to Friday, 19 September 2025 (18:00)
CERN (500/1-001)

        : Sessions
    /     : Talks
        : Breaks
15 Sept 2025
16 Sept 2025
18 Sept 2025
19 Sept 2025
AM
09:00 Introduction  
09:15
Cutting Edge AI for Offline Data Processing (until 10:45)
09:15 Neutrino Vision: AI for Neutrino Event Reconstruction - Joachim Kopp (CERN)  
09:20 Object Identification using AI/DL trained on data - Stefano Camarda (CERN)  
09:25 Transformer Model for Global PID - Christian Sonnabend (CERN, Heidelberg University (DE))  
09:30 Unsupervised Object Identification for Particle-Detector Data - Dr Thomas Poschl (CERN)  
09:35 AI/DL-based track reconstruction in non-Standard applications - Benjamin Huth (CERN)  
09:40 End-to-end track reconstruction - Noemi Calace (CERN)  
09:45 AI modeled artificial retina for global track finding - Eric Cano (CERN)  
09:50 B-field calibration through model training on tracking residual errors - Eric Cano (CERN)  
09:55 GNNs and Transformers for reconstruction tasks in CMS - Sebastian Wuchterl (CERN)  
10:00 Enhance jet energy measurements using machine learning tools for ATLAS data analysis and on-line triggering - Tancredi Carli (CERN)  
10:05 Integration and validation of ML-based particle flow (MLPF) in Phase-2 TICL reconstruction - Joosep Pata (National Institute of Chemical Physics and Biophysics (EE)) Farouk Mokhtar (Univ. of California San Diego (US)) Marco Rovere (CERN)  
10:10 Algorithms for sequential ML-based Particle Flow Reconstruction - Peter McKeown (CERN)  
10:15 Calibration of object identification algorithms using normalizing flows in CMS - Davide Valsecchi (ETH Zurich (CH))  
10:20 Design and deployment of calibrated NNs for neutral B mesons tagging - Alberto Bragagnolo (CERN)  
10:25 Event Generation and Reconstruction with the CMS HGCAL using Modern ML - Benedikt Maier (K)  
10:30 An End-to-end solution for event building at the Future Circular Collider: Machine Learning Based Track Reconstruction and Particle Flow - Dr Lena Maria Herrmann  
10:35 Large Physics Model: a foundation model for HEP data reconstructions and analysis with CMS data - Sebastian Wuchterl (CERN)  
10:40 federated learning for cross-experiment foundation models - Maurizio Pierini (CERN)  
10:45 --- Coffe Break ---
11:05
Cutting Edge AI for Offline Data Processing (until 12:45)
11:05 Improving Fast Hadronic Shower Simulation for ATLAS and Future Calorimeters - Michael Duehrssen-Debling (CERN) Nedaa Alexandra Asbah (CERN)  
11:10 Reference Datasets for AI/DL research - Andreas Salzburger (CERN)  
11:15 Simulation of hadronic showers - Anna Zaborowska (CERN)  
11:20 Data Representations for Fast Shower Simulation - Anna Zaborowska (CERN)  
11:25 Efficient energy deposit mapping in highly granular calorimeters - Anna Zaborowska (CERN)  
11:30 Extending CaloChallenge: A live benchmark for ML-based calorimeter simulations - Anna Zaborowska (CERN)  
11:35 Distillation of Diffusion models (exploration applicable to other models) - Anna Zaborowska (CERN)  
11:40 A robust validation metric for evaluating particle showers - Peter McKeown (CERN)  
11:45 CaloChallengev2 for LHCb - Michał Mazurek (National Centre for Nuclear Research (PL))  
11:50 Detector geometry-aware model - Peter McKeown (CERN)  
11:55 Data-driven Tuning of Fast Simulation Models - Peter McKeown (CERN)  
12:00 Compression of generative models for fast simulation - Peter McKeown (CERN)  
12:05 ML-based punch-through surrogate: Enhancing the realism of fast shower simulation - Anna Zaborowska (CERN)  
12:10 RICH ML Challenge for LHCb - Michał Mazurek (National Centre for Nuclear Research (PL))  
12:15 Generative ML models for RICH detectors: Investigate the gain for Cherenkov detectors - Anna Zaborowska (CERN)  
12:20 Self-Supervised Learning for Fast Detector Simulation via Generative Modeling - Dr Sofia Vallecorsa (CERN)  
12:25 FlashSim development and integration - Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)) Maurizio Pierini (CERN)  
12:30 Towards a common end-to-end flash simulation - Michał Mazurek (National Centre for Nuclear Research (PL))  
12:35 End-to-end Fast Detector Simulation and Reconstruction - Nedaa Alexandra Asbah (CERN) Michael Duehrssen-Debling (CERN)  
09:00
AI for metadata analysis (until 10:10)
09:00 Enhancing data quality control with AI in ALICE - Mr Barthelemy Von Haller (CERN)  
09:05 Automated data quality monitoring production and commissioning - Pedro Vieira De Castro Ferreira Da Silva (CERN)  
09:10 Intelligent Observability: Applying AI Techniques to the Telemetry and Operations of the LHC computing facilities - Vasco Barroso (CERN) Matteo Concas (CERN)  
09:15 AI for experiment control - Tassilo Rauschendorfer (Politecnico di Milano (IT))   (40/S2-A01 - Salle Anderson)
09:20 AI-Driven Anomaly Detection and Data Quality Strategies for the AMBER Experiment - Dr Thomas Poschl (CERN)  
09:25 AI Solutions for Infrastructure Management and Facility Operations Support - Jaroslaw Szumega (CERN EP-DT-DD, Mines ParisTech (FR))  
09:30 Trusted AI Agent Infrastructure for Repository and Library Workflows at CERN - IT-CA-IR/OSI  
09:35 Evaluation of AI Techniques for Data Centre Cooling Optimization and Resource Allocation in the LHCb Computing Infrastructure - Pierfrancesco Cifra (CERN)  
09:40 Design of a Real-Time Anomaly Detection System for LHCb Operational Logs - Benedict Kamoni Njoki (University of Nairobi (KE))  
09:45 AI-assisted Data Quality Assessment for LHCb - Titus Mombächer (University of Cincinnati (US))  
09:50 AI/DL assisted detector (parameter) optimization - Andreas Salzburger (CERN)  
09:55 AI-assisted optimization of a noble-liquid ionization calorimeter for FCC-ee - Nikiforos Nikiforou (CERN) Andreas Salzburger (CERN) Martin Aleksa (CERN)  
10:00 Let the Data Shape the Model: Towards Accessible Machine Learning - Axel Naumann (CERN) Leonardo Beltrame (Politecnico di Milano (IT)) Christine Zeh (Vienna University of Technology (AT))  
10:05 ML-Driven Orchestration of Heterogeneous Workloads on the Experiment HPC Facilities - Lubos Krcal (CERN)  
10:10 --- Coffee Break ---
10:30
AI Infrastructure for Model Training (until 11:50)
10:55 Energy-Aware Training Strategies for Sustainable Generative Modeling in Detector Simulation - Dr Sofia Vallecorsa (CERN)  
11:00 A modular ML training framework for state-of-the-art HEP tools and analysis - Sebastian Wuchterl (CERN)  
11:05 MLOps Infrastructure and End-to-End Workflows for Online LHCb Operations - Apostolos Karvelas (CERN)  
11:10 Distributed data-loading Pipelines with ROOT for large-scale ML Training - Stephan Hageboeck (CERN)  
11:15 Zero-conversion reading of HEP data for training with common ML tools - Dr Vincenzo Eduardo Padulano (CERN)  
11:20 Provisioning AI/ML tools for data scientists - Andre Sailer (CERN)  
11:25 A Filesystem View on AI Training Data in Object- and Cloud Storage - Valentin Volkl (CERN)  
11:30 Scaling out AI/ML workloads to external resources - Raulian-Ionut Chiorescu Ricardo Rocha (CERN)  
11:35 Data Preparation for Machine Learning Event Reconstruction - Lena Maria Herrmann  
11:40 Network infrastructure to support on-premises AI/ML workloads - David Gutierrez Rueda (CERN) Eric Grancher (CERN)  
11:45 itwinai: Scalable AI Training and Optimization on HPC for Science - Matteo Bunino (CERN) Dr Maria Girone (CERN)  
11:30
Infrastructure for AI Deployment (until 13:00)
11:50 Infrastructure for deployment and management of AI Agents (Agentic AI) - Ricardo Rocha (CERN)  
11:55 An experiment-agnostic bookkeeping system for trained inference models - Danilo Piparo (CERN)  
12:00 Evaluation of heterogeneous hardware solutions for AI deployment - Thorsten Wengler (CERN) Ioannis Xiotidis (CERN)  
12:05 ML Operations (MLOps) for FPGA based trigger implementations - Thorsten Wengler (CERN) Ioannis Xiotidis (CERN)  
12:10 Model registry, versioning, traceability and reproducibility - Amine Lahouel (CERN)  
12:15 AI/ML benchmark suite for hardware procurement of accelerator devices - Hannes Jakob Hansen  
12:20 Kubeflow backed by CVMFS: Efficient ML Model distribution for the Grid - Valentin Volkl (CERN)  
12:25 Efficient Integration of AI/ML workflows in HEP Frameworks (e.g., Key4hep) - Andre Sailer (CERN)  
12:30 Streamlining integration of ML inference in typical HEP analysis workflows - Dr Vincenzo Eduardo Padulano (CERN)  
12:35 Optimization of Machine Learning Inference - Sanjiban Sengupta (CERN, The University of Manchester)  
12:40 Efficient Heterogeneous Machine Learning Inference - Lukasz Michalski (Wroclaw University of Science and Technology (PL))  
12:45 Centralized inference infrastructure in Gaudi - Maarten Van Veghel (Nikhef National institute for subatomic physics (NL)) Michał Mazurek (National Centre for Nuclear Research (PL))  
12:50 Production-level ML training and deployment pipelines at LHCb - Maarten Van Veghel (Nikhef National institute for subatomic physics (NL)) Michał Mazurek (National Centre for Nuclear Research (PL)) Dr Nicole Skidmore (University of Warwick)  
12:55 Data Management for AI/DL workflows with Rucio - Martin Barisits (CERN) Mario Lassnig (CERN)  
10:00 Patter Recognition and Object Identification Discussion - Noemi Calace (CERN) Andreas Salzburger (CERN)   (40/4-C01)
09:00 Introduction - Maurizio Pierini (CERN) Dr Sofia Vallecorsa (CERN) Lorenzo Moneta (CERN)  
09:10
Cutting Edge AI for Offline Data Processing (until 10:10)
09:10 Global Event Interpretation - Maurizio Pierini (CERN)  
09:25 Pattern Recognition - Andreas Salzburger (CERN)  
09:40 Generation/Simulation - Anna Zaborowska (CERN)  
09:55 AI for HEP TH - Jacob Friedrich Finkenrath (CERN)  
10:10 --- Coffee Break ---
10:30 Optimal AI deployment for Online Data Processing - Sioni Paris Summers (CERN)  
10:50 AI for metadata analysis - Maurizio Pierini (CERN)  
11:05 Training and Education - Felice Pantaleo (CERN)  
11:15 Large Language Models-based assistants - Maurizio Pierini (CERN)  
11:30 Discussion  
11:55 Experimental Technologies - Dr Sofia Vallecorsa (CERN)  
PM
12:45 --- Lunch Break ---
14:00
Cutting Edge AI for Offline Data Processing (until 15:30)
14:00 Dreaming at CERN - Andrea Caputo (CERN)  
14:05 AI for QCD theory - Vitaly Magerya  
14:10 Charting the space of scattering amplitudes with neural optimizers - Alexander Zhiboedov (CERN)  
14:15 Machine learning trivializing flows for gauge theories - Jacob Friedrich Finkenrath (CERN)  
14:20 Building a TH group “Physics for AI” - Jacob Friedrich Finkenrath (CERN)  
14:25 Neural network quantum states for the CERN nuclear physics program - Giuliano Giacalone  
14:30 ML based global fit for gravitational wave detection with LISA - Valerie Domcke (CERN)  
14:35 Optimized and scalable statistical analysis and numerical optimization with autograd and ML frameworks - Josh Bendavid (CERN)  
14:40 Computationally efficient and smooth systematic variations from experimental response distributions with normalizing flows - Josh Bendavid (CERN)  
14:45 AI-Ready C++ code for simulation-driven analysis in HEP and beyond - Jonas Rembser (CERN)  
14:50 Modeling of Machine Induced backgrounds for LHCb - Michał Mazurek (National Centre for Nuclear Research (PL))  
14:55 Machine induced background in detector simulation - Peter McKeown (CERN)  
15:30 --- Coffe Break ---
16:00
Experimental Technologies (until 17:20)
16:00 Simplicial attention mechanism for physics objects - Sebastian Wuchterl (CERN)  
16:05 Robust and Efficient Topological Deep Learning for Event Reconstruction - Christine Zeh (Vienna University of Technology (AT))  
16:10 Accelerating AI with in-memory computing devices - Maurizio Pierini (CERN)  
16:15 Tensor Networks for Particle Physics - Maurizio Pierini (CERN) Ms Ema Puljak (The Barcelona Institute of Science and Technology (BIST) (ES))  
16:20 QML at LHCb - Dr Nicole Skidmore (University of Warwick) Jacco Andreas De Vries (Nikhef National institute for subatomic physics (NL))  
16:25 Hybrid Monte Carlo event generators - Dr Sofia Vallecorsa (CERN) Dr Michele Grossi (CERN)  
16:30 Warm-starting Variational Quantum Algorithms via Pre-Training on Classically Simulable Structures - Jogi Suda Neto (University of Alabama (US))  
13:00 --- Lunch Break ---
14:00
Optimal AI deployment for Online Data Processing (until 16:00) (40/S2-A01 - Salle Anderson)
14:00 Deployment of Machine Learning Algorithms into Hardware Trigger systems - aka MLOps - Maciej Mikolaj Glowacki (CERN)  
14:05 Improving fast inference solutions at LHCb - Maarten Van Veghel (Nikhef National institute for subatomic physics (NL))  
14:10 hls4ml and conifer - Sioni Paris Summers (CERN)  
14:15 Edge AI for Feature Extraction and Data Preprocessing on FPGAs - Dr Thomas Poschl (CERN)  
14:20 Evaluation and Integration of AMD AI Engines for low latency inference in HL-LHC Experiments - Dimitrios Danopoulos (CERN)  
14:25 pQuant - Roope Oskari Niemi  
14:30 Automation of optimization / quantization methods in the CERN MLOps offering - Amine Lahouel (CERN)  
14:35 AI/DL Long Lived Particles triggering in the Atlas Muon Spectrometer for Phase-2 HL-LHC - Davide Di Croce (CERN)  
14:40 Online Track Seeding with Edge-Classification GNNs in High Occupancy Environments - Christian Sonnabend (CERN, Heidelberg University (DE))  
14:45 Distributed Reconstruction in FPGAs and ASICs for Trigger Systems - André David (CERN)  
14:50 Common-mode noise rejection for the Phase 2 CMS HGCAL - Arne Christoph Reimers (CERN)  
14:55 AI-Enabled Custom Chips for Front-End and Back-End Electronics in HEP Experiments - Sioni Paris Summers (CERN)  
15:00 Anomaly Detection in the CMS Level 1 Trigger (Run 3 & Phase 2) - Maciej Mikolaj Glowacki (CERN)  
15:05 LLM-based Reasoning tools for front-end and back-end processing algorithm design - Maurizio Pierini (CERN)  
15:10 ML Anomaly detection Triggers with Continuous Flows and Vector Field-based algorithm - Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)) Maurizio Pierini (CERN)  
15:15 CMS: Anomaly detection and event classification with transformers for L1 Scouting on Versal AI engine - Elias Leutgeb (CERN)  
15:20 Unsupervised Anomaly Detection for the Trigger - Thorsten Wengler (CERN) Markus Elsing (CERN) Ioannis Xiotidis (CERN)  
15:25 Super resolution models for offline-like reconstruction in the scouting stream - Sebastian Wuchterl (CERN)  
15:30 ML in the CMS Phase 2 Level 1 Trigger - Sioni Paris Summers (CERN)  
15:35 Hit-based flavour tagging applications at trigger level using AI/DL - Markus Elsing (CERN) Lorenzo Santi (CERN)  
15:40 Enhancing Classical Trigger Algorithms with Machine Learning - Thorsten Wengler (CERN) Ioannis Xiotidis (CERN)  
15:45 GPU (de)compression - Jolly Chen (CERN & University of Twente (NL))  
15:50 Edge SpAIce - Sioni Paris Summers (CERN)  
16:00 --- Coffee Break ---
16:20
Large Language Models-based assistants (until 17:50)
16:20 GINO: a Grid-INtelligent Operator for Monte Carlo and Analysis - Maximiliano Puccio (CERN)  
16:25 AI Chatbot for the ALICE O² Bookkeeping System - Mr George Raduta (CERN)  
16:30 AI Assistant for ATLAS operations and beyond - Carlos Solans Sanchez (CERN)  
16:35 AI-assisted Coding Tools for LHCb - Titus Mombächer (University of Cincinnati (US))  
16:40 AI for training and documentation at LHCb - Dr Nicole Skidmore (University of Warwick)  
16:45 LHCb Analysis Chatbot - Dr Nicole Skidmore (University of Warwick)  
16:50 LLM for HEP co-pilot - Georgios Karathanasis (CERN)  
16:55 Proposal for a working group to develop AI for scientific writing at CERN - Micha Moskovic (CERN)  
17:00 Streamlining Technology Support of Distributed HEP Communities - Dr Maria Arsuaga Rios (CERN)  
17:05 Federated learning towards a common CERN chatbot - Maurizio Pierini (CERN)  
17:10 A multi-agent system for LHCb computing operations - Alexandre Franck Boyer (CERN)  
17:15 AI-Powered development tools: adoption challenges and access framework - Ismael Posada Trobo (CERN)  
17:20 Catalog and infrastructure for LLM serving and fine-tuning - Ricardo Rocha (CERN)  
17:25 Systematic Assessment of Common LLMs wrt. HEP Analysis Codes - Danilo Piparo (CERN)  
17:30 LLM compression - Maurizio Pierini (CERN)  
17:35 Operational Intelligence for Computing Operations, enabled by Generative AI - Panos Paparrigopoulos (CERN) Alessandro Di Girolamo (CERN) Dr Maria Arsuaga Rios (CERN)  
17:40
Training and Education (until 18:00)
17:40 STEAM Academy: a multi-year training backbone for Machine Learning at CERN - Felice Pantaleo (CERN)  
17:45 Courses and best practices for ML development and operations (MLOps) - Ricardo Rocha (CERN) Raulian-Ionut Chiorescu  
17:50 mPP tutorials - Maurizio Pierini (CERN)  
17:55 AI sprint at CERN - Dolores Garcia (CERN)  
12:05 Software and Hardware Infrastructure - Ricardo Rocha (CERN)  
12:20 Final Discussion