Peiyu Zhang
Hi there! 👋 I am a third-year Ph.D. student in Computer Engineering at the University of Southern California (USC), advised by Prof. Paul Bogdan. I received my B.E. in Electronic Information and Communication Engineering from Huazhong University of Science and Technology (HUST) in 2024.
My research focuses on Multi-Agent LLM Systems and Machine Learning Systems (MLSys), with interests in agentic post-training, LLM modeling, and efficiency. I am particularly interested in long-context reasoning and the latency and efficiency of multi-agent systems, as well as MLSys topics including reasoning compression and efficient operators (kernels). I have also worked on LLM agents for hardware design (spec-to-HDL generation) and graph neural networks (GNNs).
I am always open to potential collaborations, please feel free to reach out.
🔥 News
- 2026.10:  🎉 One paper accepted to EMNLP 2026!
- 2026.07:  🎉 One paper accepted to ICCAD 2026!
- 2026.07:  🎉 One paper published in npj Computational Materials!
- 2026.02:  🎉 One paper accepted to MLSys 2026!
📚 Educations
University of Southern California (USC), Los Angeles, CA
Ph.D. in Computer Engineering Aug 2024 – Present
Ph.D. in Computer Engineering Aug 2024 – Present
Huazhong University of Science and Technology (HUST), Wuhan, China
B.E. in Electronic Information and Communication Engineering Aug 2020 – Jun 2024
B.E. in Electronic Information and Communication Engineering Aug 2020 – Jun 2024
💻 Experiences
Baidu Research, Beijing, China
Machine Learning Research Intern May 2023 – Jan 2024
Machine Learning Research Intern May 2023 – Jan 2024
📑 Publications
* Equal contribution
Query-Guided RoPE Modulation for Lost-in-the-Middle Mitigation in Long-Context LLMs
arXiv preprint, 2026 (under review)
Multi-Trajectory Latent Communication for Collaborative Reasoning
Under review
Paper BibTeX Code
HyperLabel: Multi-Label Classification via Hypergraph-Based Label Correlation Modeling
arXiv preprint, 2026
@article{zhang2026hyperlabel,
title = {HyperLabel: Multi-Label Classification via Hypergraph-Based Label Correlation Modeling},
author = {Zhang, Peiyu and Ping, Heng and Kanakaris, Nikos and Zhao, Yucheng and Li, Shixuan and Yang, Wei and Xiao, Xiongye and Bogdan, Paul},
journal = {arXiv preprint arXiv:2609.32276},
year = {2026}
}
Auditing Multi-Agent LLM Reasoning Trees Outperforms Majority Vote and LLM-as-Judge
Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)
@inproceedings{yang-etal-2026-auditing,
title = {Auditing Multi-Agent {LLM} Reasoning Trees Outperforms Majority Vote and {LLM}-as-Judge},
author = {Yang, Wei and Li, Shixuan and Liu, Gengshuo and Ping, Heng and Zhang, Peiyu and Bogdan, Paul and Thomason, Jesse},
booktitle = {Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing},
month = oct,
year = {2026},
address = {Budapest, Hungary},
publisher = {Association for Computational Linguistics},
url = {https://aclanthology.org/2026.emnlp-main.2642/},
doi = {10.18653/2026.emnlp-main.2642},
pages = {60101--60118}
}
Predicting Slip Transmission Pathways Using Graph Neural Networks
npj Computational Materials, 2026
@article{jin2026predicting,
title = {Predicting Slip Transmission Pathways Using Graph Neural Networks},
author = {Jin, Yuefeng and Zhang, Peiyu and Xiao, Xiongye and Mengiste, Ezra and Li, Wenxi and Das, Amlan and Shanks, Katherine and Kasemer, Matthew and Bogdan, Paul and Bucsek, Ashley},
journal = {npj Computational Materials},
year = {2026},
doi = {10.1038/s41524-026-02218-3}
}
COEVO: Co-Evolutionary Framework for Joint Functional Correctness and PPA Optimization in LLM-Based RTL Generation
IEEE/ACM International Conference on Computer-Aided Design (ICCAD 2026)
@inproceedings{ping2026coevo,
title = {COEVO: Co-Evolutionary Framework for Joint Functional Correctness and PPA Optimization in LLM-Based RTL Generation},
author = {Ping, Heng and Zhang, Peiyu and Li, Shixuan and Yang, Wei and Cheng, Anzhe and Duan, Shukai and Zhang, Xiaole and Bogdan, Paul},
booktitle = {Proceedings of the IEEE/ACM International Conference on Computer-Aided Design (ICCAD)},
year = {2026},
doi = {10.1145/3831252.3834264}
}
VeriMoA: A Mixture-of-Agents Framework for Spec-to-HDL Generation
Conference on Machine Learning and Systems (MLSys 2026)
@inproceedings{ping2026verimoa,
title = {VeriMoA: A Mixture-of-Agents Framework for Spec-to-HDL Generation},
author = {Ping, Heng and Bhattacharjee, Arijit and Zhang, Peiyu and Li, Shixuan and Yang, Wei and Cheng, Anzhe and Zhang, Xiaole and Thomason, Jesse and Jannesari, Ali and Ahmed, Nesreen and Bogdan, Paul},
booktitle = {Proceedings of Machine Learning and Systems},
volume = {8},
pages = {1277--1290},
year = {2026}
}
HDLCoRe: A Training-Free Framework for Mitigating Hallucinations in LLM-Generated HDL
IEEE International Conference on LLM-Aided Design (LAD 2025)
@inproceedings{ping2025hdlcore,
title = {HDLCoRe: A Training-Free Framework for Mitigating Hallucinations in LLM-Generated HDL},
author = {Ping, Heng and Li, Shixuan and Zhang, Peiyu and Cheng, Anzhe and Duan, Shukai and Kanakaris, Nikos and Xiao, Xiongye and Yang, Wei and Nazarian, Shahin and Irimia, Andrei and Bogdan, Paul},
booktitle = {IEEE International Conference on LLM-Aided Design (LAD)},
year = {2025}
}
End-to-End Learning Framework for Solving Non-Markovian Optimal Control
International Conference on Machine Learning (ICML 2025)
@inproceedings{zhang2025endtoend,
title = {End-to-End Learning Framework for Solving Non-Markovian Optimal Control},
author = {Zhang, Xiaole and Zhang, Peiyu and Xiao, Xiongye and Li, Shixuan and Tzoumas, Vasileios and Gupta, Vijay and Bogdan, Paul},
booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
series = {Proceedings of Machine Learning Research},
volume = {267},
year = {2025}
}
MIHC: Multi-View Interpretable Hypergraph Neural Networks with Information Bottleneck for Chip Congestion Prediction
Advances in Neural Information Processing Systems (NeurIPS 2025)
@inproceedings{zhang2025mihc,
title = {MIHC: Multi-View Interpretable Hypergraph Neural Networks with Information Bottleneck for Chip Congestion Prediction},
author = {Zhang, Zeyue and Ping, Heng and Zhang, Peiyu and Kanakaris, Nikos and Lu, Xiaoling and Bogdan, Paul and Xiao, Xiongye},
booktitle = {Advances in Neural Information Processing Systems},
volume = {38},
pages = {88312--88342},
year = {2025}
}
A Structure-Aware Framework for Learning Device Placements on Computation Graphs
Advances in Neural Information Processing Systems (NeurIPS 2024)
@inproceedings{duan2024structure,
title = {A Structure-Aware Framework for Learning Device Placements on Computation Graphs},
author = {Duan, Shukai and Ping, Heng and Kanakaris, Nikos and Xiao, Xiongye and Kyriakis, Panagiotis and Ahmed, Nesreen K. and Zhang, Peiyu and Ma, Guixiang and Capot{\u{a}}, Mihai and Nazarian, Shahin and Willke, Theodore L. and Bogdan, Paul},
booktitle = {Advances in Neural Information Processing Systems},
volume = {37},
pages = {81748--81772},
year = {2024}
}😊 Honors and Awards
- Annenberg Graduate Fellowship, USC, 2024
- Outstanding Undergraduate Graduates, HUST, 2024
- National Olympiad in Informatics in Province(NOIP), JIANGSU