Ziyuan Nan
Papers
1
Total Citations
17
H-Index
1
About
Dr. Ziyuan Nan is a leading researcher in efficient AI systems, specializing in chiplet-based architectures and on-device inference for large language models (LLMs). His most notable contribution is the development of Cambricon-LLM, a pioneering chiplet-based hybrid architecture that enables the deployment of massive 70-billion-parameter LLMs directly on edge devices like smartphones and robotics. This work addresses the critical challenge of running single-batch, low-arithmetic-intensity computations—a task notoriously difficult for traditional hardware—while preserving model intelligence and enhancing user privacy and network resilience. With 17 citations since its 2024 publication, Cambricon-LLM has quickly become a foundational reference in the emerging field of edge AI. Dr. Nan’s research bridges the gap between cutting-edge model capabilities and practical, privacy-preserving deployment, making him a key figure in advancing efficient, real-world AI systems. His work continues to inspire innovations in hardware-software co-design for next-generation intelligent devices.
Research Focus
Key Achievements
Top Papers
- 1