Bridging the Gap: Insights from the AI & GPU Frontiers Event 

On June 17th, 2025, the Research Computing team hosted ‘AI & GPU Frontiers’ – an event which brought together 85 researchers from across the university to explore applications of AI and GPU computing in academic research. 

A Thriving Research Community 

The event showcased the diversity of AI and GPU applications across the university's research landscape. From scientific computing to creative applications, researchers are exploring new possibilities with high-performance computing resources. 

Thanks to the generous support of our industry partners NVIDIA and Dell Technologies, we were able to create a platform where researchers could share their work, learn from each other, and connect with the infrastructure that powers their discoveries. 

Conference Highlights: Voices From Our Research Community 

Researchers from across the university came together to create an engaging agenda filled with valuable insights, providing a snapshot of the broad range of scientific applications for AI and GPU computing.  

Among the highlights: 

• Dr Joseph Barker from the school of Physics and Astronomy gave us a fascinating insight into the importance of FP64 precision GPUs in theoretical physics, which could have implications in other fields too. 

• Dr Elton Vasconcelos from Leeds Omics explained how the use of GPUs is accelerating dRNA-seq data analysis for the exploration of ribosomes in the field of molecular biology. 

• Professor Alison Heppenstall and Dr Ricardo Colasanti talked about the potential for advanced computing infrastructure to transform agent-based modelling – a type of computer simulation technique used to study complex systems, and which still does not typically make use of high-performance computing (HPC) resources. 

• Dr Jongrae Kim from the School of Mechanical Engineering delivered a fascinating presentation on his use of GPU programming in Julia to verify the stability of complex dynamic systems through simulation. His work has exciting applications in the design of spacecraft. 

• Dr Sam Relton from the School of Medicine explained how his project DynAIRx aims to tackle the risks of polypharmacy through AI-driven analysis of electronic health records. It is believed the new computational approaches can predict the timing of medical events more effectively than traditional approaches. 

• Dr Donald Cummins from the School of Earth and Environment talked about how, on the DEEPVOLC project, they are using deep learning and GPU accelerated video classification for predicting volcanic eruptions and building more reliable early warning systems despite limited data. 

• Ifeanyi Chukwu introduced us to LIDA's LASER Trusted Research Environment which enables secure, GPU-accelerated data science and AI research on highly sensitive datasets and has already supported over 200 researchers across £70M of funded projects. 

Along with the many benefits AI and GPU technology bring to our research community, we were also prompted to consider some of their drawbacks. For example, Mark Cobban from Dell Technologies cautioned that applications of AI in scientific research could make flawed results appear more robust and credible and that data quality is key to mitigating these negative effects. In the coffee breaks, researchers discussed plagiarism, difficulties in correct attribution, and security issues involved in using Generative AI to support code development. 

We were also joined by representatives from NVIDIA who drew our attention to a range of specialist tools and training programmes offered by NVIDIA to support scientific research.

Key Insights: Understanding Our Research Community 

One of our primary goals was to better understand how researchers are currently using AI and GPU technologies, and what challenges they face. The insights from our participant survey of 22 respondents revealed clear patterns: 

• AI is Transforming Research Workflows: The most popular use of AI among our researchers was the use of generative AI for productivity enhancement, including the use of tools that assistwith coding and development. This represents a significant shift in how researchers approach their computational work.  
• GPU Computing Powers Scientific Discovery: Current GPU usage is predominantly focused on scientific computing applications, with specialized hardware playing a critical role in some disciplines.  
• Skills Gaps Present Opportunities: Our survey revealed that the biggest blocker to effective AI and GPU usage isn't access to technology—it's knowledge and skills. Researchers identified gaps ranging from foundational software engineering skills and effective utilization of the HPC system to advanced GPU programming techniques. 

Building Bridges Beyond STEM 

While many of our attendees came from STEM fields, there were also those interested in the creative applications of these technologies. This highlights the opportunity there to expand awareness of AI and GPU applications beyond traditional scientific computing into the arts and humanities.  

Looking Forward: Strengthening Research Computing Services 

The event reinforced the vital connection between robust HPC infrastructure and research productivity. The feedback we receive

d is already informing future directions: 

• Enhanced Training Programs: We're looking for ways we can provide specialized GPU training courses to address the specific skills gaps identified by our community. One possibility is the University enrolling in the NVIDIA Deep Learning Institute (DLI) University Ambassador Program to enable access to the DLI’s suite of workshops; 

• Improved Support Resources: We’re already working to develop better self-service resources to support users of our HPC system, and to expand its coverage to include advanced GPU and AI use cases; 

• Broader Outreach: We're exploring ways to connect with researchers across all disciplines who could benefit from Research Computing infrastructure and learning opportunities. 

A Foundation for Future Innovation 

The AI & GPU Frontiers event was more than just a networking opportunity—it was a strategic investment in understanding and supporting our research community. The insights we gathered will directly inform how we develop our services, structure our training programs, and allocate our resources. 

As we continue to build our research computing community, events like this help us understand how to better support researchers as they share knowledge, identify challenges, and explore new possibilities with modern computing infrastructure. 

We’d like to take this opportunity to thank everyone who contributed their time and knowledge to make this event a success. We're looking forward to continuing our work together.

The Research Computing Team provides high-performance computing resources and expertise to support research across the university. To learn more about our services and to stay informed about future events, visit the Advanced Research Computing website (arc.leeds.ac.uk)