Skip to content
View XiaohanZhao123's full-sized avatar
🎯
Focusing
🎯
Focusing
  • MBZUAI
  • Abu Dhabi, UAE
  • 01:04 (UTC -12:00)

Highlights

  • Pro

Block or report XiaohanZhao123

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
XiaohanZhao123/README.md

Xiaohan Zhao

Ph.D. Student @ MBZUAI · Research Intern @ Microsoft Research Asia

Agentic Data · Data-Centric AI · Dataset Pruning & Distillation · Adversarial Robustness

Email · Google Scholar


👋 About me

I am a Ph.D. student at the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), currently a research intern with the Visual Computing Group at Microsoft Research Asia (MSRA).

My current research focuses on agentic data: data and executable environments that let AI agents act. My earlier work spans data optimization, including dataset distillation and dataset pruning, as well as adversarial attacks on large vision-language models.

🔬 Research interests

  • Agentic Data: data synthesis and executable environments for computer-use and tool-using agents
  • Data-Centric AI: dataset pruning, dataset distillation, and data-efficient training
  • Adversarial Robustness: transferable black-box attacks on large vision-language models

📄 Selected publications

A full and up-to-date list is on my Google Scholar.

  • Pushing the Frontier of Black-Box LVLM Attacks via Fine-Grained Detail Targeting (M-Attack v2)
    Xiaohan Zhao, Zhaoyi Li, Yaxin Luo, Jiacheng Cui, Zhiqiang Shen
    arXiv preprint, 2026 · Paper

  • Exploring 3D Dataset Pruning
    Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, Zhiqiang Shen
    arXiv preprint, 2026 · Paper · Code

  • Covering Human Action Space for Computer Use: Data Synthesis and Benchmark
    Miaosen Zhang, Xiaohan Zhao, Zhihong Tan, et al.
    arXiv preprint, 2026 · Paper

  • A Frustratingly Simple Yet Highly Effective Attack Baseline: Over 90% Success Rate Against the Strong Black-Box Models of GPT-4.5/4o/o1 (M-Attack v1)
    Zhaoyi Li*, Xiaohan Zhao*, Dong-Dong Wu, Jiacheng Cui, Zhiqiang Shen
    arXiv preprint, 2025 · Paper · Code
    * Equal contribution.

⭐ Featured projects

  • Dive into Claude Code GitHub stars
    A systematic analysis of Claude Code and its design lessons for AI agent systems.
  • FigMirror GitHub stars
    Plots new data in the style of a reference figure and returns editable Matplotlib code.
  • M-Attack v1 GitHub stars
    Official code for a transferable black-box attack on large vision-language models.
  • 3D-Dataset-Pruning GitHub stars
    Official code for Exploring 3D Dataset Pruning, with representation-aware point-cloud selection.

💼 Experience

  • Microsoft Research Asia, Research Intern, Visual Computing Group

🎓 Education

  • Ph.D. Student, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI)
  • Bachelor's degree, Nanjing University of Information Science and Technology (NUIST)

✉️ Contact

Popular repositories Loading

  1. codex-vdd-skills codex-vdd-skills Public

    Codex skills for Verifier-Driven Development: contract-first planning, verification, and review workflows.

    Python 3

  2. Omini-Text-Detectors Omini-Text-Detectors Public

    Jupyter Notebook 1 1

  3. claude-code-devkit claude-code-devkit Public

    Claude Code dev toolkit: PR review/orchestrator daemon pipeline, safety hooks, Opus reviewer agent, defense loops. Backend-agnostic (use any Claude Code config).

    Shell 1

  4. autodiff_in_jax autodiff_in_jax Public

    some examples about auto diff in jax

  5. KernelSpike KernelSpike Public

    Jupyter Notebook

  6. node-docker-example node-docker-example Public

    JavaScript