I am working at School of Computer Science and Engineering, Southeast University, Nanjing, China. My research interests lie at the intersection of AI and systems, with a particular focus on the emerging era of large language models (LLMs). I study AI infrastructure for LLM training and inference, including data center network design, high-performance simulation, and system-level acceleration. At the same time, I explore AI for systems, leveraging the rapidly advancing capabilities of LLMs in coding and reasoning to tackle complex system optimization problems such as resource allocation, scheduling, and configuration tuning. In addition, I am interested in wireless sensing and computing, especially millimeter-wave radar–based perception for indoor environments, as well as point cloud modeling and edge–cloud collaborative intelligence.
System for AI, AI for System, Simulation, AI-driven decision-making
- March 2009 - November 2014
Ph. D student, Southeast University
- December 2011 - November 2012
Visiting Ph. D student (CSC grant), INRIA, Rennes, France
- August 2007 - February 2009
Master student, Southeast University
- August 2003 - July 2007
Undergraduate student, Southeast University
- April 2023 -
Associate Professor, Southeast University, Nanjing, China
- May 2018 - April 2023
Assistant Professor, Southeast University, Nanjing, China
- August 2016 - February 2018
Research fellow, Nanyang Technological University, Singapore, supervised by Dr. Xueyan Tang
- January 2015 - April 2016
Post-doctor fellowship, INRIA, Rennes, France, supervised by Dr. Michel Hurfin
Our research activities have been continuously supported by multiple funding sources, including governmental science and technology programs, Huawei, and defense-related agencies. The following provides a brief overview of several representative research efforts.
1. AI - Infrastructure
Our research in AI infrastructure focuses on network systems for large-scale AI model training and inference in modern data centers. We develop high-fidelity simulation platforms with AI workload–driven traffic generation to capture realistic communication patterns. Building on emerging architectures such as UB-Mesh, we investigate packet-spraying routing mechanisms that balance load while mitigating packet reordering and avoiding deadlocks. In addition, we explore discrete-event simulation acceleration techniques to enable efficient evaluation and optimization of large-scale AI training systems.
Key words:
- Data Center Network architecture
- Data Center Network simulation, Network routing
- Simulation acceleration and AI for Simulation
2. AI for System
Our research on AI for systems aims to move beyond traditional, expert-driven system design by leveraging the rapidly advancing capabilities of AI, particularly in automated coding and decision-making. We explore the use of AI to model, analyze, and solve complex optimization problems arising in system design, such as resource allocation, scheduling, and configuration tuning. By integrating learning-based approaches into the system design loop, we seek to enable more adaptive, efficient, and automated solutions that can outperform manually engineered heuristics in increasingly complex environments.
Key words:
- LLM for algorithm
- Optimization
3. Multi - agent Systems
This research focuses on leveraging reinforcement learning to enable a minimal number of agents to collaboratively accomplish area coverage and persistent target tracking, while satisfying practical constraints such as energy consumption, mobility limits, and sensing range. In addition, we investigate system-level optimizations to accelerate the reinforcement learning training process, including efficient simulation, scalable data collection, and training pipeline optimization, aiming to improve both learning efficiency and real-world applicability.
Key words:
- Reinforcement learning
- Multi - agents
- Training acceleration
4. Wireless sensing & computing towards smart homes via millimeter wave radar
Millimeter-wave radar has recently attracted significant attention due to its strong sensing capabilities and reduced privacy concerns. Our research focuses on applying millimeter-wave radar to indoor scenarios, including localization, people counting, tracking, fall detection, and gesture/posture/gait recognition. In addition, we investigate point cloud completion techniques to recover missing or sparse radar observations, improving the robustness and accuracy of downstream tasks. In this line of work, we not only design machine learning–based models for diverse applications but also address system challenges such as performance optimization on embedded platforms and edge–cloud coordination.
[1] Tongfei Shao, Zheyu Du, Chuanyou Li, Tianxing Wu, Meng Wang. Fast Human Action Recognition via Millimeter Wave Radar Point Cloud Sequences Learning. In Proc. of the 33rd ACM International Conference on Information and Knowledge Management (CIKM, CCF B), 2024.
[2] Jiangwei Shang, Kun Zhang, Zhan Zhang, Chuanyou Li, Lei Qian and Hongwei Liu. An Algorithm/Hardware Co-Optimized Method to Accelerate CNNs with Compressed Convolutional Weights on FPGA. Concurrency and Computation: Practice and Experience (CCPE), 2023, accepted.
[3] Chuanyou Li, Kun Zhang, Yifan Li, Jiangwei Shang, Xinyue Zhang, Lei Qian. ANNA: Accelerating Neural Network Accelerator through Software-Hardware Co-design for Vertical Applications in Edge Systems. Future Generation Computer Systems. 2023. (IF=7.307, JCR Q1, CCF C)
[4] Jiangwei Shang, Kun Zhang, Zhan Zhang, Chuanyou Li and Hongwei Liu, A High-Performance Convolution Block Oriented Accelerator for MBConv-Based CNNs. Integration, 2022. (CCF C)
[5] Yunbo Li, Chuanyou Li, Anne-Cécile Orgerie and Philippe Raipin. WSGP: A Window-based Streaming Graph Partitioning Approach. In Proc. of the 21th IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing (CCGrid), Melbourne, Australia, May 10-13, 2021. (CCF C)
[6] Yujian Zhang, Chuanyou Li, Fei Tong, Yuwei Xu. A variable neighborhood search algorithm for energy conscious task scheduling in heterogeneous computing systems. Concurr. Comput. Pract. Exp. 33 (24), 2021. (IF=1.831, CCF C)
[7] Zheyi Liu, Darong Lai, Chuanyou Li and Meng Wang. Feature Fusion Based Subgraph Classification for Link Prediction. In Proc. of the 29th ACM International Conference on Information and Knowledge Management (CIKM), October 19-23, Online, 2020. (CCF B)
[8] Runtian Ren, Yuqing zhu, Chuanyou Li and Xueyan Tang. Interval Job Scheduling with Machine Launch Cost. IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 31, no. 12, pp. 2776-2788, 2020. (IF=3.757, CCF A)
[9] Chuanyou Li, Xueyan Tang. On Fault-Tolerant Bin Packing for Online Resource Allocation. IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 31, no. 4, pp. 817-829, 2020. (IF=3.757, CCF A)
[10] Jiayi Yuan, Hongye Li, Meng Wang, Ruyang Liu, Chuanyou Li, Beilun Wang. An OpenCV-based Framework for Table Information Extraction. In Proc. of the11th IEEE International Conference on Knowledge Graph (ICKG) 2020: 621-628
[11] Yitong Guan, Chuanyou Li, Xueyan Tang. On Max-min Fair Resource Allocation for Distributed Job Execution. In Proc. of the 48th International Conference on Parallel Processing (ICPP), August 5-8, Kyoto, Japan, 2019. (CCF B)
[12] Chuanyou Li, Xueyan Tang. Brief announcement: Towards Fault-Tolerant Bin Packing for Online Cloud Resource Allocation. In Proc. of the 29th ACM Symposium on Parallelism in Algorithms and Architectures (SPAA), July 24-26, Washington D.C., USA, 2017. (CCF B)
[13] Chuanyou Li, Michel Hurfin, Yun Wang and Lei Yu. Towards a Restrained Use of Non-equivocation for Achieving Iterative Approximate Byzantine Consensus. In Proc. of the 30th IEEE International Parallel & Distributed Processing Symposium (IPDPS), pages 710-719, May 23-25, Chicago, USA, 2016. (CCF B)
[14] Chuanyou Li, Michel Hurfin, Yun Wang. Reputation Propagation and Updating in Mobile Ad Hoc Networks with Byzantine Failures. In Proc. of the 14th IEEE International Conference on Trust, Security and Privacy in Computing and Communications (Trustcom), pages 111-118, 2015. (CCF C)
[15] Chuanyou Li, Michel Hurfin, Yun Wang. Approximate Byzantine Consensus in Sparse, Mobile Ad-hoc Networks. Journal of Parallel and Distributed Computing (JPDC), Elsevier, 74(9), pages 2860-2871, 2014. (IF=4.542, CCF B)
[16] Chuanyou Li, Yun Wang, Michel Hurfin. Clock synchronization in Mobile Ad Hoc networks Based on An Iterative Approximate Byzantine Consensus Protocol. In Proc. of the 28th International Conference on Advanced Information Networking and Applications (AINA), pages 210-217, 2014.
[17] Chuanyou Li, Michel Hurfin, Yun Wang. Reaching Approximate Byzantine Consensus in Partially-connected Mobile Networks. In Proc. of the 26th International Symposium on Distributed Computing (DISC) - Brief announcement, pages 405-406, 2012.
[18] Chuanyou Li, Yun Wang. A Robust and Secure Overlay Storage Scheme Based on Erasure Coding. In Proc. of the 5th International Conference on Pervasive Computing and Applications (ICPCA), 2010.
[19] Chuanyou Li, Yun Wang. State Synchronization for New Member Join in Byzantine Tolerant Environment. Journal of Southeast University (Natural Science Edition), in Chinese, 40(1), 2010, pages 23-28.
2022江苏省科学技术三等奖(6/11)
招生
I warmly welcome students who have solid background on Mathematics and/or Computer Science (e.g., have some experiences in ACM/ICPC, like or good at programming, et al.). I also warmly welcome students who have some experiences in hardware design (fpga, asic) or embedded systems and would like to design cool AI systems in the edge (e.g., smart homes, intelligent cabin, et al.). If you are interested in my topics, please feel free to drop me an email. :-)
Paper Reading Group
I am very happy that a new joint paper reading group on "Machine Learning System" has started. :-)
Students who want to joint us please refer to the following website: https://seu-sigmlsys.github.io/ and feel free to contact us.
