Ziyao Wang

🎓 About Me

I am a Ph.D. student (ECE) at the University of Maryland, College Park, advised by Prof. Ang Li. Before that, I received my B.E. in Computer Engineering from Wuhan University, where I worked with Prof. Qian Wang. I have been fortunate to intern at IBM Research (Summer 2024) and Sony AI (Summer 2025), working on collaborative and efficient foundation models.

I am broadly interested in building collaborative, efficient, and trustworthy machine learning systems.

🔍 Research Interests

My research focuses on collaborative foundation models and practical efficiency, spanning:

• Federated & collaborative foundation models (federated fine-tuning, personalization, communication-efficiency)
  • NeurIPS 2024: FLoRA: Federated Fine-Tuning Large Language Models with Heterogeneous Low-Rank Adaptations
    Ziyao Wang, Zheyu Shen, Yexiao He, Guoheng Sun, Hongyi Wang, Lingjuan Lyu, Ang Li.
  • ICLR 2024: FedHyper: A Universal and Robust Learning Rate Scheduler for Federated Learning with Hypergradient Descent
    Ziyao Wang, Jianyu Wang, Ang Li.
  • ArXiv: Revisiting Federated Fine-Tuning: A Single Communication Round is Enough for Foundation Models
    Ziyao Wang, Bowei Tian, Yexiao He, Zheyu Shen, Guoheng Sun, Yuhan Liu, Luyang Liu, Meng Liu, Ang Li [PDF]
• Efficient LLM systems (parameter-efficient adaptation, decoding-time collaboration, deployment constraints)
  • ICML 2025: Speculate, then Collaborate: Fusing Knowledge of Language Models during Decoding
    Ziyao Wang, Muneeza Azmat, Ang Li, Raya Horesh, Mikhail Yurochkin.
  • ArXiv: Prada: Black-Box LLM Adaptation with Private Data on Resource-Constrained Devices
    Ziyao Wang, Yexiao He, Zheyu Shen, Yu Li, Guoheng Sun, Myungjin Lee, Ang Li. [PDF]
• Trustworthy LLM services (hidden tokens, billing, safety at system level)
  • ArXiv: Predictive Auditing of Hidden Tokens in LLM APIs via Reasoning Length Estimation
    Ziyao Wang, Guoheng Sun, Yexiao He, Zheyu Shen, Bowei Tian, Ang Li. [PDF]
  • ArXiv: Invisible Tokens, Visible Bills: The Urgent Need to Audit Hidden Operations in Opaque LLM Services (co-first)
    Guoheng Sun, Ziyao Wang, Xuandong Zhao, Bowei Tian, Zheyu Shen, Yexiao He, Jinming Xing, Ang Li. [PDF]

News and Updates

• 2025.08: 🎉 I finished my Research internship at Sony AI. We submitted a paper about visual token compression in unified models.
• 2025.06: 🎉 One paper is accepted at MobiSys 2025.
• 2025.02: 🎉 One paper is accepted at ICML 2025.
• 2024.10: 🎉 Two papers are accepted at CCS 2024. One of the papers received the Distinguished Paper Award. Congrats to all co-authors!
• 2024.09: 🎉 Two papers are accepted at NeurIPS 2024.
• 2024.08: 🎉 I finished my Research internship at IBM Research.
• 2024: 🎉 One paper is accepted at ICLR 2024.
• 2023: 🎉 One paper is accepted at EMNLP 2023.

Education

• University of Maryland, College Park — Ph.D. in Electrical and Computer Engineering (Expected 2028)
• Wuhan University — B.E. in Computer Engineering (2019)

Professional Experience

• Sony AI — Research Intern (Summer 2025)
• IBM Research — Research Intern (Summer 2024)

Selected Publications

(* = equal contribution)

2025

• [ICML] Speculate, then Collaborate: Fusing Knowledge of Language Models during Decoding
  Ziyao Wang, Muneeza Azmat, Ang Li, Raya Horesh, Mikhail Yurochkin.
• [ICLR] Towards counterfactual fairness through auxiliary variables
  Bowei Tian, Ziyao Wang, Shwai He, Wanghao Ye, Guoheng Sun, Yucong Dai, Yongkai Wu, Ang Li.
• [MobiSys] EdgeLoRA: An Efficient Multi-Tenant LLM Serving System on Edge Devices
  Zheyu Shen, Yexiao He, Ziyao Wang, Yuning Zhang, Guoheng Sun, Wanghao Ye, Ang Li.
• [ArXiv] Prada: Black-Box LLM Adaptation with Private Data on Resource-Constrained Devices
  Ziyao Wang, Yexiao He, Zheyu Shen, Yu Li, Guoheng Sun, Myungjin Lee, Ang Li. [PDF]
• [ArXiv] One Communication Round is All It Needs for Federated Fine-Tuning Foundation Models
  Ziyao Wang, Bowei Tian, Yexiao He, Zheyu Shen, Luyang Liu, Ang Li. [PDF]
• [ArXiv] Predictive Auditing of Hidden Tokens in LLM APIs via Reasoning Length Estimation
  Ziyao Wang, Guoheng Sun, Yexiao He, Zheyu Shen, Bowei Tian, Ang Li. [PDF]
• [ArXiv] Invisible Tokens, Visible Bills: The Urgent Need to Audit Hidden Operations in Opaque LLM Services (co-first)
  Guoheng Sun*, Ziyao Wang*, Xuandong Zhao, Bowei Tian, Zheyu Shen, Yexiao He, Jinming Xing, Ang Li. [PDF]
• [ArXiv] CoIn: Counting the Invisible Reasoning Tokens in Commercial Opaque LLM APIs
  Guoheng Sun, Ziyao Wang, Bowei Tian, Meng Liu, Zheyu Shen, Shwai He, Yexiao He, Wanghao Ye, Yiting Wang, Ang Li. [PDF]

2024

• [NeurIPS] FLoRA: Federated Fine-Tuning Large Language Models with Heterogeneous Low-Rank Adaptations
  Ziyao Wang, Zheyu Shen, Yexiao He, Guoheng Sun, Hongyi Wang, Lingjuan Lyu, Ang Li.
• [ICLR] FedHyper: A Universal and Robust Learning Rate Scheduler for Federated Learning with Hypergradient Descent
  Ziyao Wang, Jianyu Wang, Ang Li.
• [CCS (Distinguished Paper Award)] Moderator: Moderating Text-to-Image Diffusion Models through Fine-grained Context-based Policies
  Peiran Wang, Qiyu Li, Longxuan Yu, Ziyao Wang, Ang Li, Haojian Jin.
• [CCS] Beowulf: Mitigating Model Extraction Attacks Via Reshaping Decision Regions
  Xueluan Gong, Rubin Wei, Ziyao Wang, Yuchen Sun, Jiawen Peng, Yanjiao Chen, Qian Wang.

2023

• [EMNLP Findings] UPTON: Unattributable Authorship Text via Data Poisoning
  Ziyao Wang, Dongwon Lee, Thai Le.
• [WWW] NetGuard: Protecting Commercial Web APIs from Model Inversion Attacks using GAN-generated Fake Samples
  Xueluan Gong, Ziyao Wang, Yanjiao Chen, Qian Wang, Cong Wang, Chao Shen.

Transactions

• [TIFS] A gan-based defense framework against model inversion attacks Xueluan Gong*, Ziyao Wang*, Shuaike Li, Yanjiao Chen, Qian Wang.
• [TDSC] Kaleidoscope: Physical backdoor attacks against deep neural networks with RGB filters Xueluan Gong*, Ziyao Wang*, Yanjiao Chen, Meng Xue, Qian Wang, Chao Shen

Full list: see /publications/
Last updated: 2025-12-19