Proposed a multifidelity linear regression approach that significantly reduces model variance and improves robustness in data-scarce scenarios.

Developed multifidelity linear regression methods to enhance predictive accuracy in data-poor, high-dimensional applications, showing significant improvement on a hypersonic vehicle surface pressure prediction example.

Developed projection-enabled multifidelity neural networks to reduce computational costs for a 2D aerodynamic airfoil inverse design problem.

Leveraging convolutional neural networks for rapid and efficient aerodynamic performance evaluation, offering a faster alternative to traditional CFD solvers in turbo-machinery blade design in the early design cycle.

Student-led project to develop a hybrid rocket engine using a novel Nytrox-paraffin oxidizer and fuel combination. (Featured on the news!)