I am a PhD student in Mechanical Engineering at the University of Nevada, Reno, working at the intersection of machine learning, fluid dynamics, and scientific computing.

My research focuses on learned reduced models for complex physical systems, particularly how the formulation of the learning problem shapes what a model can subsequently generate, forecast, and control.

My current work includes:

• Generative modeling of turbulent flows
• Representation learning for multiscale physical fields
• State-to-state evolution operators for parametric PDE forecasting
• Data-driven surrogate modeling for plume evolution
• Latent-space feedback control of unsteady fluid flows

I study how diffusion models, flow matching, and latent generative models can reproduce physically meaningful turbulent flow fields, with particular interest in preserving multiscale structure during compression and generation.

I develop encoder–propagator–decoder models that learn finite-time state-to-state evolution maps for parametric dynamical systems, enabling forecasts from arbitrary reachable states and variable time horizons.

I use representation learning, clustering, and optimization to identify recurring flow regimes and dynamically accessible pathways for feedback control of unsteady fluid systems.

State-to-state latent surrogate models for transient parametric PDEs.

• Latent Evolution Operator (LEO) — finite-time state-to-state forecasting from arbitrary intermediate states, physical parameters, and forecast horizons.
• ConvLEO — wind-conditioned plume forecasting from a single observation and a user-specified time jump.

Generative modeling and representation learning for scientific data.

• DDPM vs Flow Matching on MNIST — controlled comparison of diffusion and flow matching using identical U-Net backbones.
• Spectrally Regularized Latent Flow Matching — zone-weighted log-spectral compression that recovers the dissipation-range scales MSE suppresses, carried through to unconditional turbulence generation. AI4Physics @ ICML 2026.

Reduced representations for identifying and controlling nonlinear fluid-dynamical pathways.

• Cluster-based feedback control of separated-flow transients in a learned latent space — VAE-based latent macrostates and continuous cluster-aware feedback for controlling deeply separated flow over a stalled airfoil.