Quintessa placement students, Ben Porter and Danny Bosworth, have achieved second place in a global coding competition run by the Nuclear Energy Agency (NEA), with a focus on applying large language models (LLMs) and AI techniques to the processing of risk registers.
The NEA is an intergovernmental agency that supports the safe, secure and sustainable use of nuclear energy through international collaboration and the sharing of best practices. In May, the NEA will be hosting an International Workshop in Korea, focusing on Artificial Intelligence for Nuclear Energy. Ahead of this, they hosted a coding competition for students, bringing together participants from around the world to explore how AI tools can support structured risk assessment in complex domains.
Over the course of three weeks in March, the competition had teams producing standardised, machine-readable risk registers from a diverse range of partially complete input registers. It required balancing the capabilities of generative AI with the need for accuracy and reproducibility. Ben and Danny used prompt engineering techniques to constrain an LLM into generating outputs with strict formatting requirements, while still harnessing its predictive abilities to interpret and organise information.
The competition highlighted the growing role of generative AI in supporting decision-making in domains such as risk assessment, particularly where large volumes of structured and semi-structured information must be processed consistently. Quintessa is grateful to the NEA for the opportunity to engage with this evolving area and to develop our understanding of how these tools can be applied in practice.
Ben and Danny achieved a final score of 89%, narrowly placing behind the winners with a score of 90%. Despite not winning the prize trip to present their work in Korea, both will be enjoying the improving weather in Henley and Warrington respectively before returning to university!