About me | Minji Kim

Hi, I'm Minji Kim!

I am a Ph.D. candidate in Statistics and Operations Research at UNC Chapel Hill under Prof. Vladas Pipiras.
My research focuses on modeling physical systems with computer simulation codes, using statistical methods for data-driven acceleration and physics-informed AI.

(1) How to utilize data efficiently when multiple simulation codes for the same quantity with different fidelities are available? (computational cost vs. accuracy trade-off.) I focus on estimating distributions of scarce high-fidelity outputs by leveraging cheaper model outputs through (i) nonparametric kernel density estimation with importance sampling and (ii) a parametric ensemble estimator approach, both with particular emphasis on extreme behaviors.
(2) Spatio-temporal modeling in latent space: Physical AI research in physics-informed reduced-order modeling for approximating (spatio-temporal) PDE solutions. Latent state dynamics were modeled with Neural ODEs. Used coordinate-based NNs to encode/decode the space and apply insights from 3D graphics to impose boundary conditions.

Applied these skills during internships at Google (2025), Moloco (2023), and LLNL (2022). Contributed to advancing large-scale language model research at Google, tackling real-time ad bidding challenges at Moloco and data-driven physics simulations at the national research lab.

More details are in the Research tab or my Resume.

Ph.D: Statistics and Operations Research (STOR), UNC Chapel Hill, May 2026 (expected)
M.S: Statistics, Seoul National University, Feb 2021
B.S: Statistics, Minor in Computer Science and Engineering, Seoul National University, Feb 2019