Researcher

Haojin Zhu

Haojin Zhu is a computer scientist at Shanghai Jiao Tong University whose research on network security and privacy extends to attacks on, and audits of, AI models. After earning degrees at Wuhan University and Shanghai Jiao Tong University, he completed a PhD in electrical and computer engineering at the University of Waterloo in 2009. His 2021 paper on hidden backdoors in language models received an ACM CCS Best Paper Runner-Up Award, and his 2024 paper on bit-flip attacks against DNN inference won a USENIX Security Distinguished Paper Award. In 2023 he coauthored RAI² on identity audit for AI datasets and models, and in 2025 he coauthored a study of trojaned LLM plugins at NDSS. He is a Distinguished Professor and Chairman of the GC Council at SJTU Global College.

Timeline

  1. 2002–2009

    Degrees from Wuhan to Waterloo

    He earns a B.Sc. in computer science at Wuhan University in 2002 and an M.Sc. in computer science at Shanghai Jiao Tong University in 2005. He completes a PhD in electrical and computer engineering at the University of Waterloo in 2009.

  2. 2020

    An IEEE TCSC excellence award

    In 2020, he receives the IEEE TCSC Award for Excellence in Scalable Computing (Middle Career Researcher) for his contributions to secure and scalable computing in the Internet of Things.

  3. 2021

    Hidden backdoors in language models

    In 2021, he coauthors "Hidden Backdoors in Human-Centric Language Models" with Shaofeng Li, Minhui Xue and others. The paper builds covert textual backdoor triggers for toxic comment detection, neural machine translation and question answering, and it receives an ACM CCS 2021 Best Paper Runner-Up Award.

  4. 2022–2024

    IEEE VTS Distinguished Lecturer

    From 2022 to 2024, he is a Distinguished Lecturer of the IEEE Vehicular Technology Society, giving the lecture "Security and Privacy for Artificial Intelligence of Things".

  5. 2023

    Auditing AI identity with RAI²

    At NDSS in 2023, he coauthors RAI² with Tian Dong, Minhui Xue and others, a responsible identity audit scheme for deep learning datasets and models. Owners register their datasets and models with a trusted third party that handles regulation and copyright forensics.

  6. 2024

    A Distinguished Paper on FrameFlip

    In 2024, he is a coauthor, with Shaofeng Li, Minhui Xue and others, of "Yes, One-Bit-Flip Matters!", which receives a Distinguished Paper Award at USENIX Security. The paper presents FrameFlip, an attack that depletes DNN model inference with a single bit-flip injection.

  7. 2025

    Trojaning plugins of LLMs

    In 2025, he coauthors "The Philosopher's Stone: Trojaning Plugins of Large Language Models" at NDSS with Tian Dong, Minhui Xue and others. The paper shows that an infected adapter can make an open-source LLM output adversary-defined content on specific triggers.

  8. As of October 2026

    Chairman of SJTU Global College

    As Chairman of the GC Council at Shanghai Jiao Tong University Global College, he signs a Global Degree Pathways agreement with the University of Waterloo in October 2025, alongside Waterloo's Dean of Engineering Mary Wells. In April 2026, he presides over the college's 20th anniversary conference. A Distinguished Professor, he earlier served as Vice Dean of the School of Computer Science.

Career

  1. NowShanghai Jiao Tong UniversityVice Dean

Education

  1. 2009University of WaterlooPhD
  2. 2005Shanghai Jiao Tong UniversityM.Sc.
  3. 2002Wuhan UniversityB.Sc.

Sources and credits

Photo: We have no published photo of Haojin Zhu yet. Send us a free-licensed photo.

  1. Welcome to Haojin Zhu's Home Page nsec.sjtu.edu.cn. Checked 21 October 2026.
  2. Haojin Zhu | IEEE Vehicular Technology Society vtsociety.org. Checked 21 October 2026.
  3. [2105.00164] Hidden Backdoors in Human-Centric Language Models arxiv.org. Checked 21 October 2026.
  4. RAI2: Responsible Identity Audit Governing the Artificial Intelligence - NDSS Symposium ndss-symposium.org. Checked 21 October 2026.
  5. Yes, One-Bit-Flip Matters! Universal DNN Model Inference Depletion with Runtime Code Fault Injection | USENIX usenix.org. Checked 21 October 2026.
  6. The Philosopher’s Stone: Trojaning Plugins of Large Language Models - NDSS Symposium ndss-symposium.org. Checked 21 October 2026.
  7. Haojin Zhu - School of Computer Science, Shanghai Jiao Tong University cs.sjtu.edu.cn. Checked 21 October 2026.
  8. GC signs GDP agreement with University of Waterloo | SJTU GC gc.sjtu.edu.cn. Checked 21 October 2026.
  9. GC marks 20th anniversary with high-quality development conference | SJTU GC gc.sjtu.edu.cn. Checked 21 October 2026.