Vignettes from Machine Learning in the Atomistic Sciences: Bespoke and General | Tess Schmidt

FirstPrinciples Talks presents Vignettes from Machine Learning in the Atomistic Sciences: Bespoke and General Speaker: Tess Smidt To learn more about FirstPrinciples and the AI Physicist: https://www.firstprinciples.org/ Machine learning is rapidly transforming the atomistic sciences by enabling new approaches to modeling molecules, materials, and atomic-scale systems. But despite the excitement, atomistic ML is not “push-button.” Real progress still depends on deep domain knowledge and continual evaluation. In this talk, Tess Smidt shares vignettes from research at the intersection of machine learning and atomistic science, highlighting how Euclidean neural networks (E(3)NNs) and related architectures can leverage physical principles including symmetry, hierarchy and conservation laws. By embedding these physical priors directly into neural networks, models can achieve generalizable performance even when designs appear highly bespoke. Through examples spanning interatomic potentials and connections to more traditional physical frameworks, Tess explores how physics-aware ML is reshaping the computational toolkit of atomistic modeling and the way we conceptualize modeling itself. This talk is ideal for researchers and students interested in ML for molecules and materials, equivariant neural networks, E(3)-equivariance, and physics-informed machine learning. About the speaker Tess Smidt is an Associate Professor of Electrical Engineering and Computer Science at MIT and Principal Investigator of the Atomic Architects group in the Research Laboratory of Electronics. Her research sits at the intersection of physics, geometry, and machine learning, where she develops algorithms that embed physical and geometric structure into AI systems to model and design molecules, materials, and other physical systems. Smidt is a 2025 AI2050 Early Career Fellow of Schmidt Sciences and a recipient of the Air Force Office of Scientific Research Young Investigator Research Program (YIP) Award. Before joining the MIT faculty, she was the Alvarez Postdoctoral Fellow in Computing Sciences at Lawrence Berkeley National Laboratory and a Software Engineering Intern at Google Accelerated Science, where she co-developed the first Euclidean symmetry-equivariant neural networks, architectures which naturally handle 3D geometry and geometric tensor data. She earned her SB in Physics from MIT and her PhD in Physics from the University of California, Berkeley. Subscribe for more talks on AI in science and physics-informed machine learning. #AtomisticScience #machinelearning #E3NN #EquivariantNetworks #materialsscience #computationalchemistry #aiinscience #PhysicsInformedML