Md Badrul Hasan is a Ph.D. candidate in Mechanical Engineering at the University of Maryland, Baltimore County (UMBC), working in the Computational Mechanics Laboratory under the supervision of Dr. Meilin Yu. He received his B.Sc. in Mechanical Engineering from the Bangladesh University of Engineering and Technology (BUET).

His doctoral research integrates machine and deep learning with high-fidelity simulations to develop invariance-embedded subgrid-scale models for mesoscale hurricane boundary-layer flows. His broader interests include environmental fluid dynamics, turbulence modeling, and physics-informed machine learning. He is the recipient of the 2025 AIAA Professor Kirti "Karman" Ghia Memorial Award, presented at AIAA SciTech 2025.

Accurate representation of turbulent flows remains a central challenge in computational fluid dynamics and atmospheric modeling, particularly for high-impact systems such as hurricanes and wind-energy infrastructures. Conventional sub-grid-scale (SGS) closures are predominantly dissipative, often suppressing dynamically important small-scale motions, including organized turbulence and localized energy backscatter, thereby biasing boundary-layer momentum transport and storm-intensity evolution.

This work integrates physics-based modeling with machine learning across two application domains. In hurricane boundary-layer modeling, an invariance-embedded, physics-informed machine-learning framework is used to construct spatially varying eddy-viscosity fields for mesoscale Weather Research and Forecasting (WRF) simulations, thereby improving the physical consistency of the turbulence representation.

In parallel, a physics-aware cybersecurity framework is developed for wind-turbine systems operating in real-world grid environments. The approach combines deep learning with physical constraints from turbine aerodynamics and power-generation dynamics to identify stealthy sensor anomalies, data corruption, and adversarial perturbations in SCADA streams. By embedding conservation principles, operational bounds, and turbine-specific physics into the learning pipeline, the framework aims to distinguish malicious or faulty data from normal operational variability while maintaining interpretability and compatibility with existing monitoring systems.