How can scientific structure improve machine learning performance and trust?
This includes physics-informed losses, hybrid models, domain-aware features, and evaluation strategies that go beyond purely black-box prediction.
My research sits at the intersection of computational modeling, intelligent systems, and scientific rigor. I am interested in methods that are not only accurate, but also credible, interpretable, and useful in real scientific and engineering settings.
I work on problems where simulation, data, and intelligent decision systems must interact in a principled way. This includes research in numerical simulation, machine learning, deep learning, scientific machine learning, robotics, and control. Across these areas, I am especially interested in methods that preserve structure, improve reliability, and remain meaningful for real applications.
This perspective is useful for scientific forecasting, engineering analysis, robot control, and hybrid computational systems, where purely empirical methods may be insufficient on their own.
These themes are presented as a coherent research program rather than disconnected topics.
These questions can be adapted over time as your projects, publications, and dissertation direction evolve.
This includes physics-informed losses, hybrid models, domain-aware features, and evaluation strategies that go beyond purely black-box prediction.
The goal is to connect sensing, embedded implementation, and control strategies with intelligent adaptation that remains stable and useful.
Simulation data can act as structure, supervision, prior knowledge, or benchmarking evidence for learning and analysis pipelines.
Strong academic work should also be documented through clear writing, reproducible code, teaching materials, and technical outreach.
My preferred workflow begins with a domain question, moves through modeling and implementation, and ends with evidence-based evaluation, publication-quality outputs, and communication materials that help others understand and reuse the work.