Research Focus
We sit at the intersection of theoretical chemistry and computer science, developing next-generation AI methods to solve hard physical problems.
Machine Learning for Many-Body Physics
Developing expressive, equivariant neural network architectures and generative models (like Normalizing Flows) to solve the many-electron Schrödinger equation and parameterize accurate orbital-free density functionals.
Automatic Differentiation in Quantum Chemistry
Leveraging modern AD frameworks (JAX, EnzymeAD-Rust) to compute exact derivatives of complex quantum chemical observables and energy landscapes. This enables highly efficient force field development and dynamic simulations without finite-difference errors.
Generative Potential Energy Surfaces
Creating highly accurate, surrogate models for Potential Energy Surfaces (PES) combining Gaussian processes and permutationally invariant polynomials. We aim to bridge the gap between expensive ab-initio calculations and fast molecular dynamics.
AI-Guided Materials Discovery
Utilizing Bayesian Optimization and active learning loops to efficiently explore the chemical space of complex polymers and coarse-grained systems, accelerating the discovery of novel materials with targeted thermodynamic properties.