Research Master's Student at Institut Polytechnique de Paris (IP Paris)
Focusing on 3D Vision and Computer Graphics, with an emphasis on 3D Research
I am a Master's student in Data and Artificial Intelligence at Institut Polytechnique de Paris, primarily affiliated with Télécom Paris and École Polytechnique. My research interests lie at the intersection of 3D Vision, Computer Graphics, and Computer Vision, particularly in exploring how novel deep learning techniques can enhance the rich research landscape of these fields.
Currently, my research focuses on reconstructing high-fidelity 3D geometry from diverse and challenging forms of visual input. My recent and ongoing work includes high-quality 3D surface reconstruction from freehand sketches, as well as mesh reconstruction from neural representations such as 3D Gaussian Splatting.
Beyond computer graphics and vision, I also maintain a strong interest in AI for Science. I look forward to exploring how deep learning can be adapted to assist in other scientific fields and support interdisciplinary research.
ACM Transactions on Graphics (Proc. SIGGRAPH 2026)
Paper / Demo Website / GitHub / Video
NeuralSketch2Surf: The first deep learning framework for reconstructing high-quality 3D mesh surfaces from sparse and unoriented sketches at interactive rates.
SCRIPT3D is an agentic 3D creation pipeline that translates natural-language instructions into reproducible Blender operations, producing script-backed 3D assets with persistent metadata, scene indexing, spatial placement, record-based editing, camera control, and visual verification.
TurboEdit is a real-time video editing system powered by one-step diffusion, enabling interactive control via text, reference images, and speech. Designed for low-latency streaming, it supports seamless and responsive video transformation.
Telecom Paris · Internship
Subject: Normal-Free 3D Reconstruction from Sparse VR Sketches via Voxel Occupancy Learning