The UK has invested over £1.8 billion in accelerated compute infrastructure—the specialised hardware that powers AI and big data—but lacks a coordinated national system to train the people who build and run it. Research Technology Professionals (RTPs)—the engineers, system administrators, and technical support staff who maintain these platforms—have contributed an estimated £53.3 million in value to UKRI research activities, yet their training and career development remain fragmented and under-supported. The Accelerated Compute Infrastructure Training Hub (ACIT-Hub) proposes a centralised “Hub + Spokes” model, hosted at the University of Surrey, to deliver bespoke training on accelerators, networking, storage, and orchestration software. By co-designing courses with expert RTPs and partnering with national bodies (BCS, HPC-Sig) and industry (Nvidia, AMD, Graphcore), the Hub aims to reduce training costs by 45% and training time by 30% compared to current multiple-supplier approaches. If successful, the Hub will create a professionally recognised, diverse talent pool that can deploy and maintain accelerated compute infrastructure more efficiently. This directly supports the UK’s strategic interests in AI, computer vision, and big data—fields where the country has comparative strength—and underpins the digital economy that quietly powers everything from medical diagnostics to logistics and energy grids.
View original technical description
Accelerated Compute is revolutionising many fields where the UK has strategic interests and comparative strength. The UK government and industry have invested over £1.8 billion in Accelerated Compute infrastructure with expected national economic benefits exceeding £23.7 billion by 2034. As an essential part of the Accelerated Compute ecosystem, the Research Technology Professional (RTP) community have collectively contributed over £53.3 million value to the UKRI research activities by providing high standard services in infrastructure development, operation, and research support. But while the recognition of Research Software Engineers has improved significantly over the last decade, there is a chronic lack of support and nationwide coordination to develop the RTPs across the UK who build and maintain the Accelerated Compute platforms that underpin the AI and Big Data digital economy. Our proposed Accelerated Compute Infrastructure Training Hub (ACIT-Hub) draws on the experience of the Research Infrastructure Engineers (RIEs) across the successful UKRI Tier 2 JADE consortium, to address this gap in the knowledge sharing and skills development arena. Proposing an efficient ‘Hub + Spokes’ approach the ACIT-Hub will be hosted at Surrey, a driving force in AI through its Institute in People-Centred AI (PAI) and the Centre for Vision Speech and Signal Processing (CVSSP). Together these have a world-leading track record of providing training, guiding people to reach their potential and hosting compute infrastructure for Computer Vision for over 3 decades, with CVSSP currently ranked 1st in the UK for Computer Vision and 3rd for Computer Vision and AI. The ACIT-Hub will develop efficient and sustainable training solutions, to future proof the skills development of RIEs such as SysOps, Infrastructure Engineers, DevOps, System Administrators and Technical Support, who underpin large-scale Accelerated Compute. By providing bespoke training courses and events covering accelerators, networking, storage, and orchestration and management software, highlighting both the fundamental and applied applications of the latest Accelerated Compute technologies, including when to ‘burst to cloud’. This comprehensive and agile training will be co-designed and developed by expert RTPs to ensure that it is relevant, and accessible to all. The sustainability of the Hub will be balanced with the need to make training open and accessible; the final business model will reflect this to ensure that cost is not a barrier to entry for good research practice. This ‘One-Stop’ training service will reduce 45% training cost and 30% training time compared to a traditional multiple-supplier solution. The hub recognises the diversity of backgrounds present in the RIE community and will offer translational training in essential skills and work with universities and existing accrediting bodies to provide a professionally recognised talent pool that can drive uptake in research and industry of efficient methodologies for deploying, maintaining and using accelerated compute infrastructure. It will reflect this diversity across all training and CPD events, and embed sustainability awareness at all levels. The Hub will work closely with existing UKRI DRI initiatives (the Software Sustainability Institute, STEP-UP), national bodies (BCS, HPC-Sig, N8), national infrastructure (Isambard-AI, Bede and Co-STAR), industry partners (Nvidia, AMD, Graphcore, HPE, Dell, Weka) and international initiatives (EU’s PRACE), to establish long-term collaborations, and maximise resource sharing in education and research. The Hub will enable RIEs to attend national and international conferences (CIUK, RSECon, Supercomputing and NvidiaGTC) to maximise the impact of the Hub’s training and skill development outcomes.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know