Energy-based models
Learning energy landscapes that express constraints, structure, and uncertainty in scientific systems.
Research & work
Methods for scientific systems that matter.
Current directions span structured generative models, physical learning, and robust control.
Generative modeling
Developing generative approaches for scientific discovery, design, and high-dimensional data analysis.
Physics-informed learning
Combining scientific priors with data-driven models to make predictions and controls more reliable.
Projects
Ongoing open-source contribution
DeepHyper
Contributing physics-informed neural-network benchmarks and development to an open-source hyperparameter-search framework.