Assistant Professor, UC Berkeley

Chemical Engineering and EECS

I am also a member of Berkeley AI Research (BAIR), part of the AI+Science group in EECS and the theory group in Chemical Engineering, and a faculty scientist in the Applied Mathematics and Computational Research division at LBNL.

Contact info: aditik1 dot berkeley dot edu

Research interests

I am interested in developing methods in machine learning that are driven by the distinct challenges and opportunities in the natural sciences, with particular interest in physics-inspired machine learning methods. A central theme of our work is understanding what ML models learn from scientific data, and using that understanding to develop new methods across the full arc of modeling—from model design choices and training to inference-time sampling and post-training adaptation. This includes generative modeling for physical systems, surrogate models for expensive simulations, methods for reliable prediction under distribution shift, and approaches that bridge the gap between simulation and experiment. Our work spans multi-scale dynamics, from quantum and atomistic systems to continuum fluids, and connects to numerical analysis, dynamical systems, statistical mechanics, quantum mechanics, and optimization.

Some questions we’re currently thinking about include:

  • What are neural networks actually learning from scientific data? Can we characterize and understand the physical structure they discover, rather than assuming it has to be built in?

  • When do hand-crafted priors help, and when do they hold models back?

  • What methods can we develop to better extract useful predictions from a trained model, at inference time or by adapting it for specialized downstream use?

  • How do we bridge the gap between simulation and experiment, so a model’s predictions are consistent with real measurements and not only with other simulations?

  • How do we model and predict dynamics across scales, and also transfer insights across scales?

  • What principles let us scale, train, and evaluate these models so that gains translate into reliable predictions?

See the research page for a fuller overview of these directions and examples of recent work. A full list of recent publications is available here, or on Google scholar.

Group Information

I am very fortunate to advise the following PhD students and postdocs:

Sanjeev Raja
Nithin Chalapathi
Toby Kreiman
Eric Qu
Yue Jian
Yiheng Du
Kai Nelson
Theo Sternlieb

Master’s students:

Ishan Amin

Former members:

Michael Psenka (PhD) → Research Scientist at Baseten
Ritwik Gupta (postdoc) → Assistant Professor, University of Maryland
Boyu Qie (postdoc) → Assistant Professor, Southern Institute of Science and Technology
Rasmus H⌀egh (postdoc)
Daniel Rothchild (PhD) → Research Scientist at Prescient Design (Genentech)
Danny Reidenbach (MS) → Research Scientist at NVIDIA
Nick Swenson (MEng) → Software engineer at Google

Joining the group:

  • Incoming/current UC Berkeley PhD students and prospective postdoctoral researchers: please email me directly with your research interests and CV.
  • Prospective PhD students: due to a high volume of emails, I am unable to reply to every inquiry. Please apply directly to a UC Berkeley PhD program. If you apply to Chemical Engineering or EECS, you can mention my name as a faculty of interest in your application. For EECS applicants, please choose CS: AI-SCIENCE as your primary area.
  • Undergraduate students: please fill out this application form.