Embedded Files
朱 明西
Hello! Thanks for visiting my site. I am currently an Assistant Professor at Information Technology Management group at Scheller College of Business of Georgia Institute of Technology.
I am generally interested in distributed optimization and decomposition design with GPU-accelerated implementation. This includes primal-dual and higher-order distributed optimizers to mitigate heterogeneity (e.g., ADMM, PDHG, Newton methods, preconditioning, conjugate gradients), and decomposition methods (multi-block ADMM variants, block-coordinate descent, column generation) in GPU.
In aspect of applications, much of work is aimed at Large Language Models. On the training side, I design optimizers that make pretraining and fine-tuning faster and more robust to heterogeneous data and hardware. On the post-training side, I apply the decomposition machinery to the preference data that alignment depends on: RLHF and DPO assume the humans doing the labeling behave as if maximizing a single latent utility, and my recent work builds GPU column-generation solvers that test whether real preference datasets actually satisfy that assumption, and what it costs when they do not.
I graduated from Stanford University where I am fortunate to be co-advised by Professor Yinyu Ye and Professor Haim Mendelson. I completed my undergraduate study with degrees on B.A. in English Literature and B.Ec. in Economics at Beijing Foreign Studies University, and I completed my M.A. degree on Economics at Duke University.
View my Curriculum Vitae here.
Contact me at mingxi.zhu at scheller.gatech.edu
News
October 2026: Received an Amazon Research Award to support research on primal-dual algorithms for asynchronous AI post-training on AWS Trainium. Thank you Amazon!
December 2025: Serving on the Program Committee for WISE 2025 (Workshop on Information Systems and Economics) in Nashville.
Research Projects
Managing Randomization in the Multi-Block Alternating Direction Method of Multipliers for Quadratic Optimization
Mathematical Programming Computation , volume 13, (2021), with Kresimir Mihic, Yinyu Ye.
Check our solver here ! ( https://github.com/kmihic/RACQP )
How First- and Higher-Order Primal-Dual Optimizers Accelerate Distributed Training under Data Heterogeneity
with Yinyu Ye. Best Paper award, "Order up! The Benefits of Higher-Order Optimization in Machine Learning" workshop, NeurIPS, 2022, New Orleans.
Alternative Formulations and GPU Optimization methods for Discrete-Choice Preference Learning
With Haoge Chang and Kota Saito
PD-Muon: A Primal-Dual Optimizer for LLM Pretraining under Heterogeneity
Solo. Work in progress
Is Human Feedback Rationalizable? Testing Large Scale LLM Preference
With Zeqian Li and D.J. Wu. Work in progress
Hybrid Federated Learning for LLM Fine-tuning under Heterogeneous Privacy
with Shuang Gao and Olivia Sheng. WITS 2026
Federated Learning on Algorithms, Responsibilities, and Economics (FLARE) : Integrating Federated Learning into Business Research and Applications
With Xiao Liu, Hong Guo, Heng Xu. Major Revision at Information Systems Research
Mechanism for Decision-aware Collaborative Federated Learning: A Pitfall of Shapley Values
With Meng Qi. Major Revision at Management Science,
Provision of Cybersecurity Resources to SMBs: The Investment Pitfall of Security Interdependence
With Terrence August and Marius Florin Niculescu.
Learning By Doing : The Case of Online Lending
With Haim Mendelson. Major Revision at Production and Operations Management
Design Information Disclosure under Bidder Heterogeneity in Online Advertising Auctions: Implications of Bid-Adherence Behavior
With Michelle Song.
R&R at Manufacturing & Service Operations Management
Near-Optimal Dynamic Pricing in Large Networks
With Ozan Candogan and Yuwei Luo.
Teaching
Instructor, MGT 2210, Information System & Digital Transformation (2024-2026, Spring), Scheller College of Business, Georgia Institute of Technology
Instructor, MGT 8803, Economics of AI and Machine Learning (2026, Spring), Scheller College of Business, Georgia Institute of Technology
Instructor, MGT 8803, Special Issues of Information Systems (2024, Spring), Scheller College of Business, Georgia Institute of Technology
TA, MS&E 211 X, Introduction to Optimization (2021, Fall), Engineering School, Stanford University.
TA, OIT 652, Modeling (2021, Spring), Graduate School of Business, Stanford University
TA , OIT 356, Electronic Business ( 2020-2021, Spring ), Graduate School of Business, Stanford University
Math Camp Instructor ( 2016, Fall ), Department of Economics, Duke University
Services
Reviewer for Management Science, Operations Research, Information Systems Research, Manufacturing & Service Operations Management, Production and Operations Management, MIS Quarterly, Mathematics of Operations Research, etc.
Student Advisor, MS&E Undergraduate Diversity in Research Program, MS&E Diversity Equity and Inclusion Committee, 2021 - 2022
Board Member, Stanford Graduate School Business Greater China Business Club, 2018 - 2019
Honors and Fellowships
The Institutional Venture Partners Fellowship Fund, 2020 - 2021
The David S. Tappan Jr. Fellowship Fund, 2019 - 2020
The Robert J. and Doreen D. Marshall Scholarship Fund, 2018 - 2019
George A. and Barbara Cull Jedenoff Fellowship, 2017 - 2018
Page updated
Google Sites
Report abuse