Center for Energy-efficient Computing and Applications (CECA)
School of Integrated Circuits
Peking University
Room 512, Science Building #5, 5 Yiheyuan Road, Beijing, China, 100871

I am a fourth-year Ph.D. candidate at Peking University, supervised by Prof. Guangyu Sun. My research primarily focuses on alleviating the memory wall problem through architectural and system innovations, particularly in deep learning scenarios (e.g., LLM inference and serving, large embedding model inference, and emerging algorithms). I have published as the (co-)first author in top-tier computer architecture/system conferences including ISCA (won best paper award), ASPLOS, HPCA (won best paper award), DAC, PACT.

I am expected to graduate in 2027 and am open to industry positions (especially to top-talent programs). If you are aware of suitable or well-matched positions, please free to contact me via email.

Education

• Ph.D. candidate in Integrated Science and Engineering
  • School of Integrated Circuits, Peking University, 2022-now
• B.Sc. in computer science
  • School of EECS, Peking University, 2018-2022
  • Double Major: Economics

Research Interests

• Machine Learning System
  • LLM Serving System Optimization: Enhancing LLM serving quality through optimizing request scheduling and model parallelism strategies, and implement them into LLM serving frameworks.
  • Operator Optimization on Emerging Hardware: Enhancing resource utilization on emerging AI chip products by customizing operator mapping and execution dataflow.
• Domain Specific Architecture
  • Data-centric Application Acceleration: Architecting domain-specific accelerators to alleviate memory wall issues in LLM inference, large embedding model inference, generic computation, etc.
  • 3D-DRAM-based LLM Inference Accelerator Design: Leveraging hybrid-bonding-based 3D-integration to accelerate LLM inference in a wide range of scenarios (e.g., edge-side LLM inference, cloud-level LLM serving, etc.).

Industrial Experience

• Alibaba Damo Academy
  • Research Intern (Nov. 2024 - Now)
  • Mentor: Boqiang Wu, Dimin Niu
  • Topic: (1) Architecture design for 3D-DRAM-based LLM serving accelerator. (2) Operator optimization for 3D-DRAM-based LLM serving accelerator.
• ByteDance Seed Team
  • Research Intern (Jul. 2024 - Oct. 2025)
  • Mentor: Shufan Liu
  • Topic: (1) Long-context LLM serving system optimization. (2) Performance modelling and simulation for large-scale LLM serving cluster.

Awards and Honors

Academic Awards

• ISCA Best Paper Award, 2025 (2 positions, first time in China)
• HPCA Best Paper Honorable Mention: 2025 (2 positions)
• HPCA Best Paper Award: 2023 (2 positions, first time in China)

Scholarships

• Bytedance Scholarship: 2025 (20 graduate sutdents in China/Singapore)
• China National Scholarship: 2024 (top 2%)
• President Award of Peking University: 2023, 2024 (top 2%)
• Huawei Scholarship: 2023, 2025
• Third Prize of Peking University Scholarship: 2023
• Excellent Graduate, Peking University: 2022
• Yang Xin Lotus Virtue Awards (Scholarship), Peking University: 2021
• Shenzhen Stock Exchange Scholarship, Peking University: 2020
• Founder Scholarship, Peking University: 2019
• Merit Student, Peking University: 2019, 2020, 2021 (Undergraduate), 2023, 2024 (Graduate)