PhD Candidate in Mechanical Engineering at the University of Pittsburgh (graduating December 2026), specializing in scientific machine learning for solving partial differential equations. My research advances physics-informed ML through constrained optimization — specifically, a conditionally adaptive augmented Lagrangian method (CA-ALM) for physics- and equality-constrained neural networks (PECANNs) — together with a corresponding domain-decomposition method for parallel computing that makes physics-informed learning more scalable and reliable.
- 🔭 Currently working on physics-informed ML for aerodynamic shape optimization
- 🌱 Currently revising PECANNs for the incompressible Navier–Stokes equations, trained in a fully unsupervised manner to strengthen forward-problem performance and unlock the full potential of a mesh-free, differentiable formulation — advantages that show up elsewhere: in data assimilation, inverse problems, and parametric sweeps
- 👯 Looking to collaborate on open-source physics-informed ML / scientific computing projects, and open to industry opportunities
- 💬 Ask me about PINNs, constrained optimization, CFD, or HPC (SLURM/OpenMPI/GPU)
MS in Aerospace Computational Engineering, Cranfield University · BS in Automotive Engineering, Wuhan University of Technology. Previously a computational engineer at Sun Yat-sen University, running high-fidelity CFD simulations of full-scale aircraft aerodynamics.
- Conditionally adaptive augmented Lagrangian method for physics-informed learning of forward and inverse problems — CMAME, 2026 [DOI]
- Non-overlapping, Schwarz-type domain decomposition method for PECANNs — CMAME, 2025 [DOI]
- Unsupervised simulations of incompressible flows with PECANNs — arXiv, 2026 [arXiv]
