Materials Informatics:

Materials informatics provides a data-driven lens for discovering compounds and microstructures with unprecedented performance. We develop algorithms that fuse physics-based descriptors with machine-learning models to predict properties, identify synthesis pathways, and quantify uncertainty.

Our informatics toolkit spans:

• Active learning workflows that adaptively explore vast chemical and structural spaces while minimizing expensive simulations and experiments.
• Graph and geometric deep learning approaches that capture symmetry, defects, and mesoscale features essential for accurate property predictions.
• Explainable AI frameworks that translate model predictions into actionable design rules for collaborators across materials domains.

By building open datasets and scalable software, we enable the community to accelerate discovery cycles and bridge the gap between computation and laboratory realization.