I am a Ph.D. candidate in the School of Integrated Circuits at Peking University, affiliated with the Institute for Artificial Intelligence, and advised by Prof. Bonan Yan. My research focuses on communication–computation co-optimization and hardware–software co-design for large-scale agent (LLM/RL Agent) systems. Targeting the training and deployment bottlenecks of large-scale agent systems, my work spans workload modeling, communication–computation co-optimization, and heterogeneous computing prototyping, aiming to improve the training, inference, and interaction efficiency of LLM/RL Agent systems across heterogeneous platforms such as AI accelerators, FPGAs, and multi-XPU clusters.

💻 Internships

• 2026.5 - now, ByteDance, Beijing.
• Mentor: Shuai Wang, Yinxiao Feng

• Topic: Modeling and hardware–software co-design for low-latency MoE inference clusters, analyzing topology/architecture parameters across compute, communication, and topology dimensions.

• 2024.12 - 2025.12, Alibaba Damo Academy, Beijing.
• Mentor: Di Wu, Dimin Niu
• Topic: Architecture analysis and optimization for chip-to-chip interconnect in scale-up LLM systems. Built workload-driven communication models and a hybrid cycle-accurate/event-driven simulator for system-level analysis and design space exploration.

🚀 Selected Projects

• C2C-Explorer
  A workload-driven framework for chip-to-chip interconnect modeling and design space exploration for large language model systems.
• FPGA-Gym
  An FPGA-accelerated framework for reinforcement learning environment simulation.

📝 Publications

First-author papers

• [FCCM’26] Jiayi Li, et al. “FPGA-Gym-v2: FPGA-Based RL Environment Acceleration with LLM-Assisted Onboarding” Field-Programmable Custom Computing Machines, 2026.

• [DAC’26] Jiayi Li, et al. “C2C-Explorer: An Exploration Framework for Chip-to-Chip Interconnect Architectures in LLM Cloud Computing Systems.” Design Automation Conference, 2026.

• [DATE’25] Jiayi Li, et al. “PEARL: FPGA-Based Reinforcement Learning Acceleration with Pipelined Parallel Environments.” Design, Automation and Test in Europe Conference, 2025.

• [NeurIPS-OWA’24] Jiayi Li, et al. “FPGA-Gym: An FPGA-Accelerated Reinforcement Learning Environment Simulation Framework.” Advances in Neural Information Processing Systems (Workshop), 2024.

  I have also contributed to several co-authored publications in related areas.

  🎖 Honors and Awards

• 2023, Academic Excellence Award, Peking University
• 2022, Zhishan Student Award, Southeast University
• 2021, Merit Student, Southeast University
• 2018–2022, Multiple Course Excellence Awards

  📖 Education

• 2022.09 - 2027.06 (expected), Ph.D. candidate, Peking University
• 2018.09 - 2022.06, Bachelor’s degree, Southeast University