Material design has long been driven by time intensive trial-and-error experimentation. This has become a rate-limiting step as modern challenges mount in complexity, introducing overwhelming design decisions that scale with the topic. We leverage machine learning to navigate uncharted variable space and use tools of explainable AI to interpret patterns in our models, zooming in on the underlying physical phenomena. Recent work has centered on studying nanoparticle dispersion in polymer matrices, mapping novel material behavior by Bayesian Optimization and interpretable AI tools, analyzing Gang Lab’s DNA origami crystal formations using unsupervised learning.
People: Tejus Shastry, William Marshall
