LIU Qianli’s Homepage

Hey, this is Qianli. I am a third-year Ph.D. candidate at the PEI Lab of the Hong Kong University of Science and Technology (HKUST), advised by Prof. Song Guo. My research focuses on efficient and scalable foundation model serving, with a particular interest in stateful AI systems, inference-state management, distributed MoE serving, and KV-cache management. Prior to HKUST, I received my B.Sc. in Computing with a minor in Applied Mathematics from the Hong Kong Polytechnic University (PolyU).

🔥 News

[May 2026] Our work "TwinQuant: Learnable Subspace Decomposition for 4-Bit LLM Quantization" is accepted by ICML 2026.

[Feb. 2026] Our work "KVDrive: A Holistic Multi-Tier KV Cache Management System for Long-Context LLM Inference" is accepted by SIGMOD 2026.

[Dec. 2025] Our work "DIRECTOR: Accelerating Distributed MoE Serving via Online Proactive Expert Placement" is accepted by INFOCOM 2026.

[Dec. 2025] Our work "PPAI: Enabling Personalized LLM Agent Interoperability for Collaborative Edge Intelligence" is accepted by INFOCOM 2026.

[Dec. 2024] Our work "MELL: Memory-Efficient Large Language Model Serving via Multi-GPU KV Cache Management" is accepted by INFOCOM 2025.

[Sep. 2024] Started Ph.D. in CSE@HKUST, supervised by Prof. Song Guo.

[Jun. 2024] Received B.Sc. in Computing (First Class Honours) with Applied Mathematics minor from PolyU.

📃 Publication

[1]

TwinQuant: Learnable Subspace Decomposition for 4-Bit LLM Quantization

Haodong Wang, Junjie Liu, Zicong Hong, Qianli Liu, Jian Lin, Song Guo, and Xu Chen

International Conference on Machine Learning (ICML), 2026

[2]

KVDrive: A Holistic Multi-Tier KV Cache Management System for Long-Context LLM Inference

Jian Lin, Jiazhi Mi, Zicong Hong, Haodong Wang, Qianli Liu, Haoyue Zhang, Peng Li, and Song Guo

ACM SIGMOD/PODS International Conference on Management of Data (SIGMOD), 2026

[3]

DIRECTOR: Accelerating Distributed MoE Serving via Online Proactive Expert Placement

Qianli Liu, Kaibin Guo, Zicong Hong, Peng Li, Fahao Chen, Haodong Wang, Jian Lin, and Song Guo

IEEE International Conference on Computer Communications (INFOCOM), 2026

[4]

PPAI: Enabling Personalized LLM Agent Interoperability for Collaborative Edge Intelligence

Zile Wang*, Qianli Liu*, Kaibin Guo, Haodong Wang, Jian Lin, Zicong Hong, and Song Guo (* indicates equal contribution)

IEEE International Conference on Computer Communications (INFOCOM), 2026

[5]

MELL: Memory-Efficient Large Language Model Serving via Multi-GPU KV Cache Management

Qianli Liu, Zicong Hong, Peng Li, Fahao Chen, and Song Guo

IEEE International Conference on Computer Communications (INFOCOM), 2025

👨‍🏫 Teaching Experience

COMP1021: Introduction to Computer Science(Teaching Assistant)

2025 Fall · The Hong Kong University of Science and Technology (HKUST)

COMP1021: Introduction to Computer Science(Teaching Assistant)

2025 Spring · The Hong Kong University of Science and Technology (HKUST)

COMP2011: Data Structures(Teaching Assistant)

2021 Fall · The Hong Kong Polytechnic University (PolyU)

🏅 Award

HKUST Postgraduate Studentship 2024.12.06

🎓 Academic Service

IEEE ICDCS, IEEE TPDS

Reviewer · 2025

🎤 Presentations

IEEE International Conference on Computer Communications

Oral Presentation · London, UK · May 2025

IEEE Hong Kong 6G Wireless Summit 2025

Poster · Hong Kong · Sep. 2025

CSE Research and Technology Forum (RTF) 2025

Poster · Hong Kong · Apr. 2025

🤝 Leadership & Activities

Coach · HKUST Debate Team

2025 – Present

Provide debate coaching and mentoring for team members.

Coach · PolyU Debate Team

2023 – 2024

Provided debate coaching and mentoring for team members.

President · PolyU Debate Team

2021 – 2022

Led a team of 50 members and coordinated society events and internal activities.

External Secretary · PolyU Zonta Club

2021 – 2022

Supported external engagement and social service activities promoting gender equality.