Zhonghua Jiang

About Me

Hi! I am Zhonghua Jiang, a graduated Master student from AI4GC Lab at Zhejiang University, advised by Shengyu Zhang from 2023 to 2025.

My research at AI4GC covered Efficient AI, Federated Learning, and 3D Digital Human, with a particular focus on efficient multimodal large language models, KV Cache optimization, federated model aggregation, and 3D talking digital human generation.

I am currently pursuing a Ph.D. in Artificial Intelligence at the State Key Lab of CAD&CG, College of Computer Science and Technology, Zhejiang University.

During AI4GC

During my time at AI4GC Lab, I participated in both academic research and industry collaboration projects.

I worked with Alibaba Group on 3D reconstruction and on-device deployment of large models. In the Academic AIR Program, I studied real-person digital human driving and editing using 3D Gaussian Splatting, contributing to two CCF-A conference papers(CVPR 2025, ACM-MM 2024).

I also contributed to the Zhejiang Province “Jianbing” Program, where I optimized the KV Cache of multimodal large language models for compressed deployment and faster edge-side inference. This work contributed to another two CCF-A conference papers(AAAI 2026, ACL 2025).

In addition, I worked on cloud-edge collaborative evolution of large and small models, where I explored counterfactual learning to mitigate Simpson’s paradox in model aggregation. This work led to one CCF-A conference paper(AAAI 2025).

Now

Now: Ph.D. student @ ZJU, the State Key Lab of CAD&CG.