Curriculum Vitae

Education, research, and the tools I rely on. A formatted PDF is available on request.

Education

PhD, Computational MathematicsUniversity of Southern Mississippi (in progress)

Research interests

  • Physics-informed neural networks for partial differential equations
  • 3D incompressible Navier–Stokes equations
  • Numerical analysis of scientific machine learning
  • Inverse problems in 4D-flow MRI of cerebrospinal fluid
  • Hybrid PINN / finite-volume formulations for high-Reynolds-number flow

Skills

Languages. Python (advanced), JAX, PyTorch, C/C++, MATLAB, LaTeX. Numerical methods. Finite differences, finite volumes, projection methods, spectral methods, Runge–Kutta integrators, automatic differentiation. Machine learning. PINNs, Fourier-feature networks, neural ODEs, residual adaptive sampling, NTK-balanced loss weighting. Tooling. Hydra, Weights & Biases, Docker, Git, GitHub Actions, conda, DeepXDE, NVIDIA Modulus, FEniCSx, Dedalus. HPC. Single-node and multi-GPU training (jax.pmap, torch.distributed) on NVIDIA A100 / H100.

Selected projects

PINN-NS Research Workspace (2026) Modular dissertation codebase — separates differential operators, PINN architectures, samplers, trainers, and loss formulations so each can be swapped without rewriting the others. Fully config-driven (Hydra-style YAML), with a reproducibility checklist on every experiment.

3D Navier–Stokes Beltrami Benchmark (in progress, 2026) PINN validated against the closed-form Beltrami solution (Ethier & Steinman 1994), with a classical projection-method baseline for side-by-side L^2 / divergence / wall-clock comparison.

Method of Manufactured Solutions for 3D PINNs (in progress, 2026) Adam-only vs Adam→L-BFGS optimizer study on a manufactured 3D solenoidal velocity field with analytic forcing.

Service & community

To be added.

Teaching

To be added.


Last updated: 2026-05-06.