The DESY group "Information Technology" (IT) is participating in a variety of high profile national and European projects, evaluating new and innovative IT technologies. Contemporary fundamental research on universe and matter (UM) is generating data at unprecedented rates. This drives the need for high-throughput research software and continuous improvements thereof. Typically, researchers from the respective domains have not been classically trained in computer science, which often incurs technical debts and prevents maximum computational efficiency. The overall goal of the Physics LLM consortium is to address this imbalance by utilizing and developing large language models (LLMs) for research software engineering and the accompanying concepts in research data management (RDM).

Together with project partners and stakeholders from a larger collaboration, the successful candidate will investigate the merits and demerits of LLM-based code generation frameworks in software engineering for UM-workflows, including analyses of computational performance and energy efficiency. As second major building block of the project, they will investigate and implement an LLM-assisted RDM toolkit for metadata-enriched, natural-Ianguage based accessibility.