Malte Franke

The inverse design of molecules and materials with desired properties is a grand challenge in chemistry that could unlock highly impactful technologies, from new catalysts to personalized medicine. Machine learning and generative models offer a powerful paradigm to tackle this challenge. However, current models have limited transferability across chemical space, struggle with stability and synthesizability, and often fail to incorporate the geometric and physical constraints that govern molecular behavior. My PhD project aims to address these limitations by developing new methods that integrate physical priors, multi-modality and probabilistic modeling. I will explore approaches that unify structural, energetic, and functional information to enable data-efficient and steerable generation molecules and materials- accelerating discovery in chemistry.