First-year Ph.D. Student, School of Aerospace Engineering
Tsinghua University, Beijing, China
I am a first-year Ph.D. student in the School of Aerospace Engineering at Tsinghua University, advised by Prof. Yufei Zhang. I also completed my undergraduate studies at Tsinghua University. My research lies at the intersection of scientific machine learning, computational fluid dynamics, and AI-driven engineering design. I am especially interested in combining data-driven models with reliable numerical solvers and developing human–AI collaboration paradigms for engineering design.
Looking ahead, I hope to explore physics scaling: developing scalable ways to expand and leverage scientific data in order to build foundation models for physical systems that generalize across geometries, physical regimes, and engineering tasks. My long-term goal is to move AI beyond benchmark prediction and make it a fast, reliable tool for simulation, design, and optimization in real-world engineering tasks across aerospace, the automotive industry, and other areas of advanced manufacturing.
Selected publications. My name is shown in bold.
arXiv:2608.04400, 2026. Under review.
Neural surrogate predictions provide high-quality initial guesses for Newton–Krylov iterations, improving reliability on out-of-distribution aerodynamic geometries while retaining high numerical accuracy.
Theoretical and Applied Mechanics Letters, 15(5):100602, 2025.
FoilCLIP learns a shared representation between natural-language design requirements and parametric airfoil geometries, enabling language-conditioned aerodynamic shape generation.
Worked on training paradigms for embodied foundation models, including real-robot reinforcement learning, autonomous data loops, and training efficiency. Contributed to a data flywheel that turns failed real-world trajectories discovered during model exploration into negative training examples for subsequent policy updates.
Outside research, I also enjoy a few other pursuits:
“The bone club tossed into the air by an ape turned into a spaceship.”
