Weiliang Luo’s interest in molecular science started early, sparked by a periodic table tucked into the back of a dictionary. What began as curiosity—wondering how matter is arranged and transformed—quickly deepened through hands-on exposure to simple experiments and vivid chemical reactions that felt almost magical.
Today, that curiosity drives his research at MIT, where he works at the intersection of quantum chemistry, machine learning, and scientific software. His primary focus is Enerzyme, a project centered on developing next-generation neural network potential (NNP) methods for enzymatic catalysis. The goal is to make AI-driven molecular simulations more accessible, robust, and scalable for complex biochemical systems. To do this, he combines modular, physics-inspired model design with large-scale quantum chemistry, transfer learning from biochemical data, and automated active learning workflows with uncertainty quantification.
Building Enerzyme has meant tackling both scientific and engineering challenges. Early versions of the framework were powerful but difficult to maintain, relying on fragile scripts and inefficient GPU usage. Through his MolSSI Software Fellowship, Weiliang worked with mentors—Drs. Benjamin Pritchard, Taylor Barnes, and Jessica Nash—to redesign the system from the ground up. By implementing best practices for multi-GPU PyTorch jobs on Slurm clusters and restructuring key workflows, he significantly improved performance and stability. He also reworked the NEB pipeline to better handle convergence issues, turning it into a more automated and reliable process.
These improvements went beyond speed. Standardizing data structures and configuration handling reduced user error and improved reproducibility, helping transform Enerzyme from a personal research tool into shared infrastructure used across his group.
The longer-term vision for Enerzyme is to become an open, extensible platform—accessible to experimental biochemists, flexible for developers, and aligned with rapid advances in atomistic machine learning. By accelerating simulations of enzyme catalysis, particularly in challenging systems like metalloenzymes, Weiliang hopes to enable new insights into complex biochemical processes. Ultimately, this work could support mechanistic studies, enzyme engineering for sustainable chemistry, and rational drug discovery.
The MolSSI fellowship has also given him the freedom to explore ambitious ideas while connecting with a broader community of computational scientists. These interactions have strengthened his interest in building software that is not only technically sound, but widely useful.
Looking ahead, he is open to paths in academia, industry, or startups, with a consistent goal: advancing molecular science through better computational tools. He is also motivated to mentor others as AI continues to reshape the field.
Outside the lab, Weiliang channels his creativity into the arts. He directs large-scale cultural performances and experiments with incorporating generative AI into storytelling and design. He also enjoys singing and arranging a cappella music, drawn to the same collaborative energy that defines his scientific work.
If you are interested in more of his work, please visit his GitHub and LinkedIn pages
