Published several significant projects including DeepSeek-VL (a prestigious multimodal language model), DeepSeek LLM (a prestigious open-source large language model), DreamCraft3D (hierarchical 3D generation with bootstrapped diffusion prior), etc. Additionally, the work “Bringing Old Photos Back to Life” was listed as one of the top 30 AI advancements in 2020 by the renowned AI media louisbouchard.ai.
Research Experience
Currently a ZJU 100 Young Professor (Ph.D. supervisor) at Zhejiang University. Previously served as a senior researcher at Visual Computing Group of Microsoft Research Asia (MSRA) and AI research scientist at DeepSeek.
Education
Received Ph.D. degree from the Department of Electronic and Computer Engineering at Hong Kong University of Science and Technology (HKUST) in 2019; received Bachelor's degree in Engineering from Zhejiang University in 2013.
Background
Research interests include 2D/3D content creation, virtual human modeling, multimodal models, and embodied intelligence. Contributions to the field of content generation include the high-quality image translation CoCosNet series (with CoCosNet v2 being a CVPR 2021 Best Paper nominee), the industry’s first text-to-image generation diffusion model VQ-Diffusion, the first high-quality 3D diffusion generation model Rodin, the 3D generation technology DreamCraft 3D, and the well-known open-source multimodal large model DeepSeek-VL.
Miscellany
Open to welcoming self-motivated PhD candidates, Master’s and Bachelor’s students, postdocs, and research assistants. Also looking for research collaboration with industry and research labs.