Haizhong Zheng

Postdoctoral Researcher @ CMU
Ph.D. @ UMich
Email: haizhonz@andrew.cmu.edu

haizhong.jpeg

I am a postdoctoral researcher at InfiniAI Lab advised by Prof. Beidi Chen at Carnegie Mellon University. I am also a member of Catalyst Group at CMU. I received my Ph.D. degree from University of Michigan and was advised by Prof. Atul Prakash. Before that, I obtained my B.S. and M.S. degree from Shanghai Jiao Tong University, where I was advised by Prof. Haojin Zhu.

Research Interests: Efficient RL for LLMs; ML system; Agent; ML Security.

News

Jan 2026 Three papers (M2PO, Jackpot, and OPPO) accepted to ICLR 2026.
Dec 2025 RLBoost accepted to NSDI 2026.
Nov 2025 WAVE accepted to ASPLOS 2026.
Oct 2025 Gave an invited talk at Amazon SF AI Lab.
Jul 2025 Plato accepted to COLM 2025.

Selected Publications

* indicates equal contribution.

Preprint

  1. Preprint 2025
    When "Correct" Is Not Safe: Can We Trust Functionally Correct Patches Generated by Code Agents?
    Yibo Peng* , James Song* , Lei Li* , Xinyu Yang , Mihai Christodorescu , Ravi Mangal , Corina Pasareanu , Haizhong Zheng , and Beidi Chen
    arXiv Blog Code

Conference

  1. ICLR 2026
    Prosperity before Collapse: How Far Can Off-Policy RL Reach with Stale Data on LLMs?
    Haizhong Zheng , Jiawei Zhao , and Beidi Chen
    arXiv Blog Code
  2. ICLR 2026
    Jackpot: Align Actor-Policy Distribution for Scalable and Stable RL for LLM
    Zhuoming Chen* , Hongyi Liu* , Yang Zhou* , Haizhong Zheng , and Beidi Chen
    arXiv Code
  3. ICLR 2026
    OPPO: Accelerating PPO-based RLHF via Pipeline Overlap
    Kaizhuo Yan* , YingJie Yu* , Yifan Yu , Haizhong Zheng , and Fan Lai
    arXiv
  4. NSDI 2026
    RLBoost: Harvesting Preemptible Resources for Cost-Efficient Reinforcement Learning on LLMs
    Yongji Wu* , Xueshen Liu* , Haizhong Zheng , Juncheng Gu , Beidi Chen , Z. Morley Mao , Arvind Krishnamurthy , and Ion Stoica
    arXiv Code
  5. ASPLOS 2026
    WAVE: Leveraging Architecture Observation for Privacy-Preserving Model Oversight
    Haoxuan Xu* , Chen Gong* , Beijie Liu* , Haizhong Zheng , Beidi Chen , and Mengyuan Li
  6. NeurIPS 2025
    Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rollouts
    Haizhong Zheng , Yang Zhou , Brian R. Bartoldson , Bhavya Kailkhura , Fan Lai , Jiawei Zhao , and Beidi Chen
    arXiv Blog Code
  7. NeurIPS 2025
    Kinetics: Rethinking Test-Time Scaling Laws
    Ranajoy Sadhukhan* , Zhuoming Chen* , Haizhong Zheng , Yang Zhou , Emma Strubell , and Beidi Chen
    arXiv Blog Code
  8. COLM 2025
    Plato: Plan to Efficiently Decode for Large Language Model Inference
    Shuowei Jin* , Xueshen Liu* , Yongji Wu* , Haizhong Zheng , Qingzhao Zhang , Atul Prakash , Matthew Lentz , Danyang Zhuo , Feng Qian , and Z. Morley Mao
    arXiv
  9. ICLR 2025
    ELFS: Label-Free Coreset Selection with Proxy Training Dynamics
    Haizhong Zheng* , Elisa Tsai* , Yifu Lu , Jiachen Sun , Brian R. Bartoldson , Bhavya Kailkhura , and Atul Prakash
    arXiv
  10. WWW 2025
    Harmful Terms and Where to Find Them: Measuring and Modeling Unfavorable Financial Terms and Conditions in Shopping Websites at Scale
    Elisa Tsai , Neal Mangaokar , Boyuan Zheng , Haizhong Zheng , and Atul Prakash
    Oral
    arXiv
  11. NeurIPS 2024
    Learn To be Efficient: Build Structured Sparsity in Large Language Models
    Haizhong Zheng , Xiaoyan Bai , Xueshen Liu , Z. Morley Mao , Beidi Chen , Fan Lai , and Atul Prakash
    Spotlight
    arXiv
  12. ECCV 2024
    Leveraging Hierarchical Feature Sharing for Efficient Dataset Condensation
    Haizhong Zheng , Jiachen Sun , Shutong Wu , Bhavya Kailkhura , Z. Morley Mao , Chaowei Xiao , and Atul Prakash
    arXiv
  13. ICLR 2024
    CALICO: Self-Supervised Camera-LiDAR Contrastive Pre-training for BEV Perception
    Jiachen Sun , Haizhong Zheng , Qingzhao Zhang , Atul Prakash , Z. Morley Mao , and Chaowei Xiao
    arXiv
  14. ICLR 2023
    Coverage-centric Coreset Selection for High Pruning Rates
    Haizhong Zheng , Rui Liu , Fan Lai , and Atul Prakash
    arXiv Code
  15. CVPR 2020
    Efficient Adversarial Training with Transferable Adversarial Examples
    Haizhong Zheng , Ziqi Zhang , Juncheng Gu , Honglak Lee , and Atul Prakash
    arXiv Code
  16. NDSS 2018
    Smoke Screener or Straight Shooter: Detecting Elite Sybil Attacks in User-Review Social Networks
    Haizhong Zheng , Minhui Xue , Hao Lu , Shuang Hao , Haojin Zhu , Xiaohui Liang , and Keith Ross
    arXiv

Work Experience

Research Intern, Lawrence Livermore National Laboratory (LLNL), Livermore, CA May 2023 - Aug. 2023

Applied Scientist Intern, Amazon Web Services (AWS), Inc., Seattle, WA May 2021 - Aug. 2021