Mingkai Zheng

PhD student

Department of ECE @ Rutgers

Biography

Hi, there! I am Mingkai Zheng, a PhD student from Rutgers University, NJ. I am fortunate to receive supervision from Prof. Zhao Zhang. Most of my research works focus on improving the training efficiency of large language models (LLMs).

Before that, I was a research engineer at Texas Advanced Computing Center (TACC). I received my Master of Engineering degree from the Department of Electrical and Computer Engineering at Cornell University and my Bachelor of Engineering degree from University of Liverpool and Xi’an Jiaotong-Liverpool University.

Interests

High Performance Computing (HPC)
Large Language Model (LLM)
Deep Learning (DL)

Education

Ph.D. in Electrical and Computer Engineering, 2024 - 2027 (expected)
Rutgers University
M.Eng. in Electrical and Computer Engineering, 2021 - 2022
Cornell University
B.Eng. in Electrical Engineering, 2016 - 2020
University of Liverpool
B.Eng. in Electrical Engineering, 2016 - 2020
Xi'an Jiaotong-Liverpool University

Internship

Research Scientist Intern

GenBio AI

May 2025 – Nov 2025 Palo Alto, CA

Research Engineer

Texas Advanced Computing Center

Jan 2023 – Jan 2024 Austin, TX

Publications

Quickly discover relevant content by filtering publications.

Junlin Chen, Daize Dong, Huanwei Di, Haolong Jia, Jiawei Wu, Haotian Xie, Mingkai Zheng, Yang Li, Leshang Chen, Huishu Wang, Eric P. Xing, Hongyi Wang (2026). Broken Symmetry in BF16 Attention: Why FlashAttention Gradients Blow Up Late in Training. arXiv.

PDF Cite arXiv

Haotian Xie, Junlin Chen, Mingkai Zheng, Yifan Zhu, Minu Mathew, Max Burnette, Yadu Babuji, Volodymyr Kindratenko, Shivaram Venkataraman, Kyle Chard, Ian Foster, Zhao Zhang (2026). Diamond Agent: Agentic Control of Federated HPC Resources as a Service. arXiv.

PDF Cite arXiv

Mingkai Zheng, Fangru Linghu, Sikan Li, Marty Charles Kandes, Niall Gaffney, Ian Foster, Zhao Zhang (2026). A Comparative Study on LLM-enabled HPC User Support. in PEARC 2026.

Cite DOI

Haotian Xie, Junlin Chen, Mingkai Zheng, Lishan Yang, Zhao Zhang (2026). PHOENIX: Resilient LLM Training with Hot-Swapping via Zero-Overhead Checkpoint. arXiv.

PDF Cite arXiv

Mingkai Zheng, Junlin Chen, Haotian Xie, Zhao Zhang (2026). SCAPE: Accurate and Efficient LLM Training with Extreme Sparse Communication. arXiv.

PDF Cite arXiv

See all publications