Multistate Protein
Studying protein AI across connected biological states through MuSProt, with MusBench for evaluating how models transfer knowledge and remain consistent between states.
Selected Projects
Research projects in multimodal AI, medical imaging, and protein modelling.
Self-Driving Lab for Metal-Binding Protein Design
An agentic closed-loop laboratory that generates, screens, verifies, and iteratively improves selective and stable metal-binding proteins.
AI-driven anticancer peptide de novo design
Developing AI methods to generate novel anticancer peptides while jointly considering activity, selectivity, safety, and developability.
Multimodal AI for Parkinson's Disease
Learning from genetics, biomarkers, environment, and clinical examinations to study Parkinson's disease mechanisms and progression.
Foundation Models for Neuropathic Pain
Data-efficient multimodal learning for fMRI-based drug-response prediction.