CV

Education, research experience, publications, honors, teaching, and technical skills.

Contact Information

Name Zewei Deng
Professional Title Ph.D. Student in Industrial and Systems Engineering
Email deng0379@umn.edu

Professional Summary

Ph.D. student developing statistical and coding-theoretic methods for trustworthy generative AI, LLM watermarking, provenance, and uncertainty-aware learning.

Experience

  • 2026 - Present

    Minneapolis, MN

    Graduate Research Assistant
    University of Minnesota Twin Cities
    Graduate Research Assistant since Summer 2026. Advisor: Prof. Liyan Xie
    • Develop statistical and coding-theoretic methods for provenance, edit detection, and integrity verification of watermarked LLM outputs.
    • Led an ECC-based post-generation edit detection project accepted to EMNLP 2026 Findings, spanning method design, implementation, experiments, visualization, and manuscript preparation.
    • Extended the framework to local integrity checking, using error-correcting structure to verify where and how watermarked outputs have been modified.
    • Study watermarkability across language models and methods for improving compatibility with logit-based watermarks.
    • Develop uncertainty-aware sequential and agentic generative models, robust inference for delayed point processes, and physics-informed statistical learning methods.
  • Sep 2025 - May 2026

    Minneapolis, MN

    Teaching Assistant
    University of Minnesota Twin Cities
  • 2024 - 2024

    Chicago, IL

    Research Assistant
    Chicago Booth School of Business, University of Chicago
    • Worked with Prof. Dacheng Xiu on financial-text forecasting using BERT, LLaMA, and T5 representations with regression models for return and debt-related prediction.
  • 2024 - 2024

    Chicago, IL

    Machine Learning Capstone
    Alliant Credit Union / UChicago Data Science Institute
    • Contributed to a client-facing machine-learning model development project with Alliant Credit Union.

Academic Interests

Trustworthy Generative AI: LLM watermarking, provenance, statistical learning, uncertainty quantification, temporal point processes, distributionally robust optimization

Education

  • 2025 - Present

    Minneapolis, MN

    Ph.D.
    University of Minnesota Twin Cities
    Industrial and Systems Engineering
    • Expected graduation: Dec 2028
    • Advisor: Prof. Liyan Xie
  • 2023 - 2024

    Chicago, IL

    M.S.
    University of Chicago
    Applied Data Science
  • 2019 - 2023

    Urbana-Champaign, IL

    B.S.
    University of Illinois Urbana-Champaign
    Statistics
    • Minors in Computer Science and Economics
    • Highest Distinction in Statistics

Honors and Awards

  • 2023
    Highest Distinction in Statistics
    University of Illinois Urbana-Champaign
  • 2020
    Dean's List
    University of Illinois Urbana-Champaign

    Fall 2020; Spring 2021; Fall 2021; Spring 2022; Fall 2022; Spring 2023

  • 2021
    Honorable Mention
    COMAP Mathematical Contest in Modeling

Skills

Programming: Python, R, SQL, Git/GitHub, LaTeX
Machine Learning and Scientific Computing: PyTorch, NumPy, pandas, scikit-learn
Research Methods: LLM watermarking, statistical learning, temporal point processes, Wasserstein DRO, hypothesis testing