Zhenan Fan (范喆楠)

I am currently a Staff Research Engineer at the Huawei Vancouver Research Center, where I focus on optimizing large language model (LLM) inference and deployment on Huawei’s CloudMatrix SuperPod (华为云超节点). My recent work centers on scaling LLM serving for SuperPod-scale infrastructure, addressing the challenges that arise when running mixture-of-experts (MoE) models such as DeepSeek, Kimi, and Qwen on hundreds of interconnected NPUs. We recently presented some of this work in our technical report, xDeepServe: LLM Serving on SuperPod-Scale Infrastructure.

Prior to my current work on LLMs, I contributed to the design and development of OptVerse(天筹求解器), Huawei’s in-house large-scale optimization solver, focusing on algorithmic innovations and system-level integration for real-world optimization tasks in the cloud.

I obtained my Ph.D. in Computer Science from The University of British Columbia, under the supervision of Prof. Michael P. Friedlander. My dissertation, Duality in Structured and Federated Optimization: Theory and Applications, developed new theoretical and algorithmic insights into large-scale structured and federated optimization, with applications in machine learning, data mining, and signal processing.

Education

PhD, University of British Columbia, Computer Science (2022)

MS, University of British Columbia, Computer Science (2019)

BSc, University of Toronto, Mathematics (2017)