About Me
Muhammad Risyad Hasyim is a scientist and engineer working at the intersection of artificial intelligence and the physical sciences. As a postdoctoral fellow at the Simons Center for Computational Physical Chemistry at NYU, he builds machine learning models, runs large-scale computer simulations, and applies them to key problems in chemical and condensed matter physics. Most recently, he led the electrolytes effort for Open Molecules 2025 (OMol25), a collaboration with Meta FAIR to build one of the largest AI training datasets for chemistry.
His research aims to predict and control molecular responses to environmental conditions and external fields by connecting physics across scales. To this end, he combines theory and machine learning, from electronic structure to molecular motion and collective behavior. This vision connects his work in molecular polaritonics, machine learning force fields, generative modeling and sampling, and glassy dynamics.
Dr. Hasyim completed his Ph.D. at the Department of Chemical & Biomolecular Engineering @ UC - Berkeley working with Prof. Kranthi Mandadapu. He attended Penn State for undergraduate studies and performed research under Prof. Ramakrishnan Rajagopalan (focusing on electrochemical capacitors) and Prof. Michael T. Lanagan (focusing on composite solid electrolytes).
Announcements
10/16/2026: Non-thermal aging of supercooled liquids in optical cavities has been accepted in Nature Communications. Updated research pages cover molecular polaritonics, machine learning force fields, and generative modeling and sampling, with original figures and recent results.
04/27/2026: DenSNet is now on arXiv. The collaboration uses learned electron densities to predict molecular dynamics and infrared spectra, with transfer from short to longer polythiophene chains. Read more in the research summary.
03/20/2026: Non-thermal aging of supercooled liquids in optical cavities is now up on arXiv.
03/05/2026: Work on vibrational polaritonics and mixed quantum-classical dynamics for cavity-modified chemistry is now published in the Journal of Chemical Physics, as part of an invited contribution to the special topics series on simulating excited states in condensed phases.
05/13/2025: Collaboration with Meta, LBNL, and many other institutions on generating the largest DFT dataset for machine learning interatomic potentials is now up on arXiv. See press for more details.