Yunfan ZHANG

I am a Research Scientist at the Alibaba-NTU ANGEL CorpLab, working on LLMs and 3D computer vision. My current research spans domain-specific LLMs for clinical applications (GraphRAG, KAG, multi-turn diagnostic dialogue), inference efficiency, and 3D generative pipelines. I hold a Ph.D. in Computer Science from Nanyang Technological University.

Before returning to academia, I spent eight years as a Senior Engineer at Marvell and Skyworks, designing ultra-high-speed, low-power ICs with several granted U.S. patents. I enjoy building systems that bridge hardware-aware engineering with modern AI research, and I’m always open to connecting with collaborators in AI, computer vision, and electronic design.

news

Nov 09, 2025	Passed Ph.D. oral defence at NTU.
Jun 01, 2025	Paper “Style-Consistent 3D Indoor Scene Synthesis with Decoupled Objects” accepted to IJCNN 2025.
Jan 15, 2025	Started as Research Scientist at the Alibaba-NTU Global e-Sustainability CorpLab (ANGEL).
Apr 15, 2024	Two papers on image aesthetics assessment accepted at ICASSP 2024 and ICME 2024.

selected publications

Style-Consistent 3D Indoor Scene Synthesis with Decoupled Objects

Yunfan Zhang, Hong Huang , Zhiwei Xiong , and 4 more authors

In International Joint Conference on Neural Networks (IJCNN) , 2025

Bib

@inproceedings{zhang2025styleconsistent,
  title = {Style-Consistent 3D Indoor Scene Synthesis with Decoupled Objects},
  author = {Zhang, Yunfan and Huang, Hong and Xiong, Zhiwei and Shen, Zhiqi and Lin, Guosheng and Wang, Hao and Vun, Nicholas},
  booktitle = {International Joint Conference on Neural Networks (IJCNN)},
  year = {2025}
}

InsTex: Indoor Scenes Stylized Texture Synthesis

Yunfan Zhang, Zhiwei Xiong , Zhiqi Shen , and 3 more authors

arXiv preprint arXiv:2501.13969, 2025

arXiv Bib

@article{zhang2025instex,
  title = {InsTex: Indoor Scenes Stylized Texture Synthesis},
  author = {Zhang, Yunfan and Xiong, Zhiwei and Shen, Zhiqi and Lin, Guosheng and Wang, Hao and Vun, Nicholas},
  journal = {arXiv preprint arXiv:2501.13969},
  year = {2025},
}

StarNet: Style-aware 3D Point Cloud Generation

Yunfan Zhang, Hao Wang , Guosheng Lin , and 3 more authors

arXiv preprint arXiv:2303.15805, 2023

arXiv Bib

@article{zhang2023starnet,
  title = {StarNet: Style-aware 3D Point Cloud Generation},
  author = {Zhang, Yunfan and Wang, Hao and Lin, Guosheng and Vun, Nicholas and Shen, Zhiqi and Miao, Chunyan},
  journal = {arXiv preprint arXiv:2303.15805},
  year = {2023},
}