A new generation of generative AI techniques allow for the computational design of proteins. This includes particular (and yet unknown) proteins, that specifically recognize and bind to other molecules. However, the generated protein structures reproduce desired properties only with a low probability. Thus, still a large amount of time and money is spent on experimental  testing the designed proteins. In order to improve the efficacy of AI-based protein design, we will extend and improve state-of-the art virtual pipelines for protein design by adapting their internal mathematical structure to our specific demands. We will build on existing tools (such as RFDiffusion) to generate nanobody (mini-antibody)  proteins, that recognize biomarkers or pathogens. This will be done in close interaction between the intern, a mathematician, hosted in AMCS in the group Krause, and the laboratory of Stefan Arold. In the latter, final protein designs will be produced and characterized for testing and validation. Their resulting combination of mathematical and biological competences will give rise to a protein development cycle, which will be specifically tailored towards applications. As an international partner, the well known Zuse-Institute (ZIB) in Berlin (Germany) will join this project. It is foreseen that the intern will stay for additional six months at ZIB after his internship in KAUST, continuing on the same topic.