Embedded Files
Assistant Professor of Mathematics at North Carolina State University
I am an applied and computational mathematician focused on understanding why modern machine learning works and leveraging those insights to build reliable, theoretically principled algorithms for physical and engineering systems—including fluids, thermodynamics, plasma physics, and digital twins.
My research centers on “practical theory”: designing simple, elegant algorithms that solve real-world problems. For me, proving theorems is a means to a greater end—deepening our collective understanding, asking uncharted questions, and developing creative computational tools that make AI trustworthy for science.
About
I'm an Assistant Professor in the Department of Mathematics at NC State University, where I lead a research group at the intersection of computational mathematics, machine learning, and scientific computing. Before NC State, I held positions at Brown University and Korea Advanced Institute of Science and Technology (KAIST), and I completed my PhD in mathematics at The Ohio State University. My work pairs rigorous theory with scientific machine learning that solves real physical problems in fluids, plasma physics, and thermodynamics. I care about building AI that is not just accurate but trustworthy, and about training the next generation of researchers to do the same.
News
August 2026 — Appointed Associate Editor, Advances in Computational Science and Engineering (ACSE).
July 2026 — WGFINNs published in Computer Methods in Applied Mechanics and Engineering. A weak-formulation version of GENERIC-informed neural networks that stays accurate on noisy measurements where strong-form methods break down — the regime real sensor data actually lives in. Joint work with Lawrence Livermore National Laboratory.
May 2026 — Received the John Griggs Faculty Award for Outstanding Research, NC State Department of Mathematics, recognizing outstanding research by tenure-track faculty.
April 2026 — Neural-operator element method published in Nature Computational Science. NOEM makes finite element simulation faster and more scalable by replacing repeated local solves with reusable, pre-trained neural operators — a drop-in acceleration path for existing FEM pipelines.
November 2025 — New industry project with LG Electronics (PI). "Data-free Operator Learning for Real-Time Turbulent CFD Simulations in Air-Conditioning Applications" — a second consecutive award extending the 2024–25 collaboration on physics-informed AI for fast CFD in home appliances.
Page updated
Google Sites
Report abuse