PhD student in the group of Boris Kozinsky at Harvard SEAS.
I work on molecular dynamics for materials science, with a focus on developing fast-but-accurate machine learning potentials. In particular, I work on profiling, optimization and GPU acceleration.
Highlights:
- FLARE (mir-group/flare), a fast and Kokkos-accelerated machine learning potential for extreme-scale simulations
- Phoebe (mir-group/phoebe), a fast and Kokkos-accelerated solver Boltzmann transport code
- Allegro (mir-group/pair_allegro), a scalable equivariant neural network potential with state-of-the-art accuracy
Contact: [email protected]