AI for Science Day

Date: 10 April 2025

Time: 10:00-17:00

Location: Scott Logic Theatre, Department of Computer Science

This workshop will bring AI experts from NVIDIA, Dell and Durham together to discuss joint success stories, to identify challenge of the future, and to identify collaboration pathways and new projects. Participation is open for all Durham staff, PhD students and RSEs. Guests from other UK institutions are welcome.

We expect the workshop to provide a platform to speak about the latest technology trends in AI and how they affect our scientific workflows, ambitions and plans. The workshop will exclusively focus on the science side, yet provide participants, if they want to, to have 1:1 conversations with NVIDIA and Dell specialists about particular technical solutions and products. Our overall intention is to offer a stage for new insight through AI but also for knowledge exchange around upcoming AI trends, computational facilities in the NE and internationally, and networking around AI technology.

Registration: Click here to register

Registration deadline: 4th April 15:00

Agenda:

09:00 - 10:00	Registration and coffee (Vis Lab, MCS 1022)
10:00 - 10:15	Welcome Remarks
10:15 - 10:45	Andy Grant: NVIDIA Keynote
10:45 - 11:15	Ira Shokar: AI for Physics-Informed Modelling: Advancing Weather, Climate, and Engineering
11:15 - 11:45	Ulrik Beierholm: t.b.c.
11:45 - 12:15	Paul Graham and Steve Davey: Enabling and training by NVIDIA
12:15 - 13:15	Lunch (Vis Lab, MCS 1022)
13:15-13:30	Dell Welcome Note
13:30-14:00	Tobias Weinzierl: Machines and machine access in Durham, the UK and the European Union
14:00-14:30	Rita Fernandes Neves: Generative AI in NVDIA - from NIMs to Blueprints
14:30-15:00	Coffee
15:00 - 15:30	Tobias Weinzierl, Alan Real Eamonn Bell: Durham University Digital Research Infrastructure Initiatives
15:30 - 16:00	Q&A and close
16:00 - 17:00	Meet the experts – 1:1 conversations with NVIDIA and Dell experts (Vis Lab, MCS 1022)

Details:

Ira Shoka (NVIDIA) will speak on AI for Physics-Informed Modelling: Advancing Weather, Climate, and Engineering. AI is accelerating computational fluid dynamics across a variety of applications, from high-fidelity aerodynamic simulations to modelling future weather and climate. By producing surrogate models using deep learning, we are able to reduce computational cost by orders of magnitude—enabling workflows that were previously computationally intractable. This presentation explores the advancements and applications of physics-informed AI within NVIDIA, including Earth-2, a digital twin of the planet, and PhysicsNemo, a library for building, training, and fine-tuning physics AI models at scale.

Tobias Weinzierl is the Director of Durham's Institute for Data Science and also chairs the EuroHPC JU's AI Resource Allocation panel. In this talk, he will give a brief tour de force through the AI equipment at Durham University with an outlook which resources researchers can use (free of charge) within the European Union.

Rita Fernandes Neves is a Senior Solutions Architect for Gen AI at NVIDIA AI Enterprise. Her presentation explores the advancements and applications of generative AI software within NVIDIA. From NVIDIA Inference Microservices to comprehensive AI blueprints, the goal is to detail the entire stack to provide insights and practical examples of how to use it in your AI projects.

Tobias Weinzierl, Alan Real and Eamonn Bell are the PIs behind multiple upskilling and research grants funded under the umbrella of the UK’s Digital Research Infrastructure (DRI). In this brief overview talk, they will present the agenda behind these grants and how they can help researchers to work more effectively and efficiently with the Digital Research Infrastructure in Durham and the UK.