Mingyuan (William) Zhang
Lifelong learner and builder.
I am currently working in the tech industry, building reliable agent systems that combine LLM reasoning, reinforcement learning, tool use, harnesses, evaluation, and scalable infrastructure.
I obtained my Ph.D. in Computer and Information Science from the University of Pennsylvania, where I worked on statistical machine learning for complex classification problems.
Prior to my Ph.D., I completed my undergraduate studies at the University of Michigan, where I earned degrees in Honors Mathematics, Honors Statistics, Computer Science, and Data Science.
🎓 Education
2024 — Ph.D. in Computer and Information Science
University of Pennsylvania
Dissertation: Statistical Machine Learning for Complex Classification Problems
2018 — B.S. in Honors Mathematics, Honors Statistics, Computer Science, and Data Science
University of Michigan
🔬 Research Interests
My current research focuses on large language models and agents.
My doctoral research focused on statistical machine learning for complex classification problems, with particular emphasis on:
- Learning from Noisy Labels: Designing algorithms that can learn good classifiers despite noisy training data, for both multiclass and multi-label learning problems
- Multi-Label Classification & Label Ranking: Developing effective learning algorithms for multi-label losses and various label ranking metrics
- Non-decomposable Performance Measures: Optimizing complex performance measures (F1, AUC, etc.) that cannot be decomposed per example
- Weakly Supervised Learning: Learning from missing or partial labels, and transfer learning
🏆 Honors & Awards
Top Reviewer, NeurIPS, 2024
Outstanding Achievement in Mathematics Awards, University of Michigan, 2017, 2018
James B. Angell Scholar, University of Michigan, 2015, 2017, 2018
William J. Branstrom Freshman Prize, University of Michigan, 2014
University Honors, University of Michigan, 2013 – 2018
👥 Professional Service
Journal Reviewer: JMLR, IEEE TPAMI
Conference Reviewer: NeurIPS (2021–2024), ICLR (2022–2023), AISTATS (2024)