INFN/CERN on LLM as a Service

Wednesday 18 Nov 2026, 09:00 → 17:00 Europe/Zurich

31/S-028 (CERN)

INFN CERN Exchange on LLM as a Service

09:00 → 09:20 1. Welcome and objectives

Introductions, expectations for the visit, and key questions each organisation would like to address.

09:20 → 09:40 2. INFN and CERN service overviews

Each team presents its current activities, planned services, target users, and main challenges.

09:40 → 10:00 3. Use cases and requirements

Research assistance, coding, document search, and internal knowledge assistants; expected demand and performance requirements.

10:00 → 10:20 4. Infrastructure and model serving

GPU resources, deployment architecture, model selection, inference frameworks, scaling, and resource sharing.

10:20 → 11:50 Visit: CERN computing infrastructure tour (guide: Paulo)

Subject to availability and visitor-access arrangements.

12:30 → 14:00 Lunch

14:00 → 14:20 5. Access, security, and data governance

Authentication, permissions, handling sensitive information, logging, and conditions for using external versus locally hosted models.

14:20 → 14:40 6. Service operation and evaluation

Monitoring, availability, user support, quotas, usage accounting, and evaluating model quality, latency, and efficiency.

14:40 → 15:00 7. Demonstrations and technical exchange

Short demonstrations from both teams, practical lessons learned, and discussion of retrieval-augmented generation (RAG) and API integration.

15:00 → 15:20 8. Potential collaboration and next steps

Identify shared interests, possible joint tests or benchmarks, contact persons, and follow-up actions.