Me
Zhenxin Ai
Research Assistant, HKUST(GZ)
aikunkun34 [at] gmail [dot] com

About

I am a Research Assistant at the Hong Kong University of Science and Technology (Guangzhou), working under the supervision of Assistant Professor. Haiyun He. I received my B.Eng. in Artificial Intelligence from Jiangxi University of Science and Technologye, where I was advised by Prof. Huilan Luo. During my senior year, I was a visiting student at Westlake University under the supervision of Assistant Professor. Huan Wang. My research interests span computer vision, multi-modal learning, and watermarking for large language models. I am always open to related collaborations—if you are interested in discussing potential projects, feel free to contact me via email.

Interests

  • Computer Vision
  • Multi-modal Learning
  • Watermarking for Large Language Models
  • 3D Scene Reconstruction

Education

Jiangxi University of Science and Technology
  • B.Eng in Artificial Intelligence, Sep. 2021 - Jun. 2025
  • Advisor: Prof. Huilan Luo

Experiences

Feb, 2025 - Jul, 2025
Algorithm Intern @ Zhipu AI
Focus: Large Language Model Inference and Hallucination
Jul, 2024 - Jan, 2025
Visiting Student @ Westlake University

Publications

PASA
PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attacks
Zhenxin Ai, Haiyun He
ICML 2026
[Paper] [Page] [Code]
Niagara
Niagara: Normal-Integrated Geometric Affine Field for Scene Reconstruction from a Single View
Xianzu Wu*, Zhenxin Ai*, Harry Yang, Ser-Nam Lim, Jun Liu, Huan Wang
TCSVT 2025
[Paper] [Page] [Code]
LiteSalNet
A Lightweight Multi-Stream Framework for Salient Object Detection in Optical Remote Sensing
Zhenxin Ai, Huilan Luo, Jianqin Wang
TGRS 2025
[Paper] [Page] [Code]

GitHub

Official implementation of PASA: A principled embedding-space watermarking approach for LLM-generated text under semantic-invariant attacks.
Official PyTorch implementation of Niagara: Normal-Integrated Geometric Affine Field for Single-View Scene Reconstruction.
A PyTorch implementation of LiteSalNet for Remote Sensing Salient Object Detection.

Selected Honors & Awards

Academic Services

  • Coming Soon...