I am a first-year Ph.D. student at McGill & Mila supervised by Prof. Derek Nowrouzezahrai. I was a visiting student at Westlake University supervised by Prof. Tailin Wu. Before that, I was a research intern at Chinese Academy of Sciences supervised by Prof. Lin Gao and Prof. Yu-Kun Lai.
EqCollide was accepted to the KDD 2026 AI4Science Track as an Oral presentation.
Received the FRQNT Master's Scholarship, ranked first in my research track.
Started my Ph.D. studies at McGill University and Mila.
graphics and machine learning
EqCollide is an equivariant neural-field simulator for deformable-object collisions, evolving latent control points with collision-aware message passing and decoding continuous velocity fields for stable 2D and 3D rollouts.
SeG-Gaussian uses segmentation guidance to optimize 3D Gaussian representations for high-quality novel view synthesis.
NeuralSlice is a novel approach to representing 3D shapes with flexible topology in a pure explicit fashion, which represents a 3D shape as the intersection of a 4D deformable tetrahedral mesh and a learnable 4D hyperplane.
Sparse Dynamic Convolution (SD-Conv) naturally integrates sparse convolution and dynamic convolution such that it can inherit the advantage of dynamic mechanism and sparsity.
