I am Kaiyuan Liu (刘开元). I received my BE degree from Henan University, Henan, China, in 2023, and my MS degree in Data Science from the University of Macau, Macau, China, in 2025. I am currently working as a research assistant at the State Key Laboratory of IoTSC, University of Macau. My research interests center on MLsys, on-device LLM inference, and Edge computing. I am passionate about exploring how intelligent systems can be optimized for real-world applications and advancing the efficiency of AI on resource-constrained devices.
The 46th IEEE Real-Time Systems Symposium (RTSS’25,
Author： 🌟Kaiyuan Liu, Lizi Zhang, Chengzhong Xu, Li Li
IEEE Transactions on Parallel and Distributed Systems (TPDS’25,
Author： 🌟Kaiyuan Liu, Xiaobo Zhou, Li Li
University of Macau, Jan. 2025 – May. 2025
This project focuses on developing a collaborative inference framework using idle neighboring devices to address memory and compute constraints in edge-device LLM deployment. We employ two-level optimization: inter-device coordination and intra-device resource utilization.
University of Macau, Jun. 2024 – Dec. 2024
I am investigating ways to optimize LLM inference on mobile devices, addressing issues like high energy consumption and long inference times. The framework incorporates multi-dimensional optimization to balance performance, real-time efficiency, and energy consumption.
Henan University, Apr. 2021 – May. 2022
I led the development of a Chinese Sign Language recognition model that achieved 88% accuracy for 300 signs. This project won the National Innovation and Entrepreneurship Training award and contributed to one software copyright.
- Outstanding Graduate, Henan University (Top 10%), 2023
- Henan University Scholarship (Top 10%), 2022 & 2023
- Provincial 1st Prize, CUMCM, Henan Province, 2021
- Email: [email protected]
- GitHub: 56dot1
- Google Scholar: Kaiyuan Liu
Feel free to explore my repositories, or contact me if you're interested in collaboration, research discussions, or AI-related topics!