Di Mou

AI Infrastructure Engineer
Alibaba Group
935522618 [at] qq.com

About Me

I am Di Mou, an AI infrastructure engineer working on efficient training and inference for large language models.

Professional Experience

Alibaba Group · AI Infrastructure Engineer · 2025 – Present

I focus on low-precision quantization, LLM inference optimization, and high-performance operator/kernel optimization.

Baidu · Research and Development Intern · 2024

I worked on distributed LLM training on Kunlun AI accelerators.

Education

2022–2025 · Chongqing University — Master’s degree; research on AI accelerators and hardware-software co-design
2018–2022 · Fuzhou University — Bachelor’s degree

Research Interests

AI Infrastructure: LLM systems, efficient inference, and agentic infrastructure
GPU Systems: kernel generation, profiling, and performance optimization
Computer Architecture: AI accelerators and hardware-software co-design

Selected Projects

Atrex Kernel Agent — an open-source agent for GPU kernel implementation, profiling, analysis, and iterative optimization

Publications

arXiv Dissecting Outlier Dynamics in LLM NVFP4 Pretraining

Peijie Dong, Ruibo Fan, Yuechen Tao, Di Mou, Wenhu Hu, Zhenheng Tang, Yinghao Yu, Jiamang Wang, Wenbo Su, Guodong Yang, Liping Zhang, Xiaowen Chu, Baochun Li, Bo Li

arXiv preprint arXiv:2602.02047, 2026.

Paper PDF
IEEE TPDS SC-CGRA: An Energy-Efficient CGRA Using Stochastic Computing

Di Mou, Bo Wang, Dajiang Liu

IEEE Transactions on Parallel and Distributed Systems (IEEE TPDS), vol. 35, no. 11, pp. 2023–2038, 2024. (CCF-A)

Paper DOI
ICCAD DISC: Exploiting Data Parallelism of Non-Stencil Computations on CGRAs via Dynamic Iteration Scheduling

Yue Liang, Di Mou, Dajiang Liu

43rd IEEE/ACM International Conference on Computer-Aided Design (ICCAD), pp. 1–9, 2024. (CCF-B)

Paper DOI
DAC DARIC: A Data Reuse-Friendly CGRA for Parallel Data Access via Elastic FIFOs

Dajiang Liu, Di Mou, Rong Zhu, Yan Zhuang, Jiaxing Shang, Jiang Zhong, Shouyi Yin

60th ACM/IEEE Design Automation Conference (DAC), pp. 1–6, 2023. (CCF-A)

Paper DOI

Powered by Jekyll and Minimal Light theme.