Skip to content

Isambard Summit 2026

Draft programme

This programme is a draft and is subject to change. Speakers, titles, timings, and abstracts may be updated before the event.

The Isambard Summit 2026 is a two-day event bringing together researchers, engineers, and industry partners to share progress and discuss challenges in AI and high-performance computing. Click any talk title to read the full abstract in the detailed programme.

Rosalind Franklin Room - Day 1

Session 1: Welcome and Keynote | 10:00–11:15

Time Speaker Affiliation Title
10:00 Simon McIntosh-Smith Bristol Centre for Supercomputing Welcome and introduction to Bristol Centre for Supercomputing
10:10 TBC UK Government TBC
10:20 Fred Manby Iambic Building the NeuralPLexer4 Co-Folding Model
11:05 Richard Gilham Bristol Centre for Supercomputing Conference tracks and dinner

Refreshments | 11:15–11:45

Session 2: Large Language Models and AI Research | 11:45–13:00

Time Speaker Affiliation Title
11:45 Pontus Stenetorp University College London UK-LLM after three years: Reflections and a road map for the future
12:10 Aleksej Zelezniak King's College London AI for Synthetic Genome Design
12:35 Huw Day University of Bristol Understanding Partitioned Learning Dynamics

Lunch | 13:00–14:00

Session 3: AI for Health | 14:00–15:15

Time Speaker Affiliation Title
14:00 Aldo Faisal Imperial College London Nightingale-AI
14:25 Jon Lees University of Bristol AI as a Bridge Across Cellular Scales
14:50 Gregory Verghese PharosAI, King's College London Towards Transparent AI in Computational Pathology: Multimodal Concept Learning for Clinical AI

Refreshments | 15:15–15:45

Session 4: AI for Advanced Materials | 15:45–17:00

Time Speaker Affiliation Title
15:45 Matthew Foulkes Imperial College London Neural Wavefunctions for Materials Chemistry and Physics
16:10 Gabor Csanyi University of Cambridge MACE force field models for the periodic table
16:35 Panel SME focus panel - title to follow

Close of Day 1 | 17:15


Rosalind Franklin Room- Day 2

Session 5: Keynote | 09:00–10:15

Time Speaker Affiliation Title
09:00 TBC Bristol Centre for Supercomputing Welcome
09:05 Jeffrey S. Vetter Oak Ridge National Laboratory Keynote — TBC
09:50 David Topping University of Manchester Partnerships at Scale: HPC, AI and the Future of Environmental Decision-Making

Refreshments | 10:15–10:45

Session 6: AI Security | 10:45–12:00

Time Speaker Affiliation Title
10:45 Jason Gwartz AI Security Institute An Introduction to the AI Security Institute and AI Safety Research on Isambard
11:10 Yalli Du King's College London Evaluating the cooperative behaviour of systems of generative agents
11:35 Sid Black AI Security Institute Auditing games for sandbagging detection

Lunch | 12:00–13:00

Session 7: AI Research and Closing Address | 13:00–15:00

Time Speaker Affiliation Title
13:00 Zilin Wang University of Oxford Learning to Drive in New Cities Without Human Demonstrations
13:25 Eghbal Rahimikia University of Manchester Re(Visiting) Time Series Foundation Models in Finance
13:50 Bidipta Sarkar University of Oxford Evolution Strategies at the Hyperscale
14:15 Eltayeb Ahmed and Anya Sims University of Oxford Reinforcement Learning for Mid-Training on Unstructured Text
14:40 Closing Address

Close of Day 2 | 15:00


Annex- Day 1: SME Focus in Partnership with NVIDIA

Running in parallel with the main programme.

Part 1 | 11:45–13:00

Time Speaker Affiliation Title
11:45 TBC Bristol Centre for Supercomputing Welcome, Bristol Centre for Supercomputing introduction and aims for the day
12:00 Jessica Driscoll NVIDIA Accelerating AI Innovation: The NVIDIA Inception Program and the Supercomputing Ecosystem
12:20 Jamil Appa Zenotech TBC

Lunch | 13:00–14:00

Part 2 | 14:00–15:15

Time Speaker Affiliation Title
14:00 Wasil Rezk BeyondMath TBC
14:25 Karin Sevegnani NVIDIA From Infrastructure to Impact: Sovereign AI Development on Isambard AI

Annex- Day 2: Parallel Sessions

Running in parallel with the main programme.

Parallel Session 1 | 10:45–12:00

Time Speaker Affiliation Title
10:45 Tim Santos Graphcore What changes when you go from 1 node to 'lots'
11:10 Ian Johnson HPE Securing Containerised AI Kubernetes Workloads on Isambard AI with Slingshot
11:35 Duncan Roweth HPE AI Inference with NVIDIA Dynamo on HPE's Slingshot Network-based Systems

Lunch | 12:00–13:00

Parallel Session 2 | 13:00–15:00

Time Speaker Affiliation Title
13:00 Pili Mayora and Dan Lenton AI Security Institute AISI Research Platform: Isambard technical workflows
13:25 Sadaf Alam Bristol Centre for Supercomputing AIRR Status and AI Data Facility Update
13:50 David Africa AI Security Institute Consistency Training
14:45 Reconvene in main hall

Workshops- Day 1

Workshop 1: Research Data Mobility in the Age of AI | 14:00–15:15 | Learning Room 3

Organisers: James Womack and Laura Shemilt

In computational research and AI workflows, data moves through complex pipelines across multiple digital research infrastructures (DRIs). Varied access control mechanisms, network configuration, and software on DRIs may inhibit automated data movement. This workshop explores how federated identity systems and shared protocols can enable seamless data mobility across DRIs. First, we present our National Federated Compute Services (NFCS) NetworkPlus project, which surveys operator and end-user requirements for data movement and evaluates data centre APIs with federated identity as solutions. Second, a panel discussion with NFCS team members and guests, with Q&A on our project and discussion of the data mobility challenges in the age of AI.

Panel:

  • Chris Edsall (University of Cambridge)
  • James McClung (VAST Data)
  • Lynsey Smart (University of Bristol)
  • Pontus Stenetorp (University College London)
  • Stig Telfer (StackHPC Ltd)
  • James C. Womack (Bristol Centre for Supercomputing)

Workshop 2: Interactive AlphaFold Tutorial | 15:45–17:00 | Learning Room 3

Organiser: Wahab Kawafi

This hands-on workshop introduces AlphaFold for protein structure prediction using JupyterHub notebooks on the Isambard AI GPU cluster. Participants will explore protein–ligand binding design for drug discovery through interactive exercises and visualisations. Attendees will gain practical skills in leveraging high-performance computing resources for computational biology. No prior AlphaFold experience required.

Workshop materials

Workshops- Day 2

Workshop 3: DigiEngBio Accelerator | 10:45–12:00 | Design West — The Assembly (off site)

AI and digital technologies are transforming Engineering Biology (EngBio), speeding up data analysis, improving design processes, and enabling predictive models that drive precise innovation. DigiEngBio will accelerate innovation at the interface of AI and EngBio by leveraging Isambard AI to build an integrated, accessible, digital EngBio design platform that will:

  • Dramatically speed up research in protein, polymer, and whole-genome design
  • Democratise access to advanced AI tools for the academic and industrial EngBio community
  • Enable scalable, data-driven innovation unlocking applications across sectors

Workshop 4: Distributed PyTorch Tutorial | 10:45–12:00 | Learning Room 2

Organiser: Anna Price

This workshop introduces how to dispatch a distributed training job on Isambard AI, bridging the gap between training on a single node and distributing jobs across multiple nodes. Learning objectives:

  • Learn to launch a distributed PyTorch job
  • Understand how Slurm, MPI, and NCCL interact
  • Understand how MPI and NCCL are dispatched and the modules required for them
  • Combine the above to make use of the high-speed network (Slingshot)

Distributed training tutorial | Workshop materials

Workshop 5: TREs with Kubernetes on Isambard AI | 10:45–12:00 | Learning Room 3

Organiser: Jake Watson

Research projects working with sensitive health data — such as records from UK Biobank — must operate within Trusted Research Environments (TREs) that meet strict governance requirements. Running these workloads on shared AI infrastructure presents a challenge: how do you maintain the isolation and security a TRE demands on a multi-tenant system? The FRIDGE project, funded through DARE UK, addresses this by using Kubernetes to provide a secure, isolated deployment platform that meets TRE requirements while giving researchers access to large-scale compute.

Workshop 6: Porting Containers between DRI | 13:00–14:15 | Learning Room 2

Organiser: Richard Gilham

Do you use containers in your research? Do you need to move your research between different systems? Would a managed container registry improve your research? This workshop provides an opportunity for a facilitated discussion on experiences, challenges, and opportunities, in support of the National Federated Compute Service (NFCS) Federated Container Registry project.

Time Speaker Affiliation Title
13:00 Chris Edsall University of Cambridge Introducing the NFCS Federated AI Container Project
13:10 TBC TBC
13:30 Discussion

Workshop 9: Cybersecurity — What's it all about? | 13:00–14:15 | Learning Room 3

Organiser: Thomas Green

Cybersecurity in digital research infrastructure is central to a modern interconnected world where infrastructure can be exposed to many threats, from nation-state actors to inquisitive individuals. This workshop will explore and encourage discussion on the methods that can be used to protect infrastructure — exploring threats, methods to evaluate protection, and what steps can be taken. Using the criteria in the NCSC CAF framework as a guide, the workshop will help answer questions such as "What are the challenges to assess larger digital research infrastructure assets?". At the end of the workshop, attendees will have discussed and understood some of the most recent cybersecurity topics.


Posters

Presenter Affiliation Title
Yi Liu University of Bristol MR-KG: A knowledge graph of Mendelian randomization evidence powered by large language models
Lon Barfield University of Bristol Delivering and managing Digital Research Infrastructure: Lessons from the Joiner platform
Liam Berrisford University of Exeter Enabling LFRic on Isambard 3: A Reproducible AArch64 Software Stack Using Spack
Tanya Kushwahaa Cardiff University From Prototype to Production: Agentic AI Tools for Natural Language-Driven HPC Management
James Brock University of Bristol Forest-Chat: Adapting Vision-Language Agents for Interactive Forest Change Analysis
Babatunde Oyewo University of Cambridge Automated Encryption of Kubernetes Block Storage for Trusted Research Environments (TREs)