CV | Zheyuan Xiao

Zheyuan Xiao

Title

LLM Researcher and Software Engineer

Location

Auckland, New Zealand

xiaozheyuan72@gmail.com

Website

zxia545.github.io

Summary

M.S. candidate in Computer Science at The University of Texas at Austin researching reliable LLM evaluation, data-centric post-training, social simulation, and agentic systems. Alongside research, I build automated verification systems for firmware and cloud-connected products at Resideo.

Education

Jan 2023 – Present Austin, Texas, USA

The University of Texas at Austin · M.S. in Computer Science

GPA: 4.0/4.0

Jul 2018 – Dec 2021 Auckland, New Zealand

The University of Auckland · B.E. (Hons.) in Software Engineering

First Class Honours · GPA: 7.029/9

Research Experience

Jan 2026 – Present Remote

Allen Institute for AI (Ai2) · Research Collaborator

Aug 2025 – Mar 2026 Remote

Microsoft Research Asia · Research Collaborator, TalentScope Copilot

LLM-driven autonomous web agents for evidence-grounded academic talent discovery.

Built a browser agent that searches, navigates, collects evidence, and synthesizes candidate profiles.
Developed multi-agent orchestration, provenance tracking, and evaluation with AutoGen, Playwright, TypeScript, and Python.

Dec 2024 – Jun 2025 Remote

Microsoft Research Asia · Research Collaborator, LLM Knowledge Inheritance

Studied how domain knowledge is retained and transferred during language-model fine-tuning.

Built data and training pipelines for domain-specific QA across multiple knowledge areas.
Fine-tuned Qwen2.5, Llama 3.1, and Mistral 7B models with LoRA and supervised fine-tuning on multi-node 8×A100 clusters.
The resulting work was published at the NeurIPS 2025 Main Conference.

Feb 2024 – Feb 2026 Remote

HKUST(GZ) · Research Assistant

LLM preference evaluation, post-training data selection, and social simulation.

Investigated desirability and information mass as drivers of length bias in LLM-based evaluation.
Contributed to work published at Findings of EMNLP 2025 and accepted to EMNLP 2026 Main Conference.

Industry Experience

Mar 2025 – Present Auckland, New Zealand

Resideo (Honeywell Home) · Software Engineer II

Built an end-to-end automated test system spanning GitHub Actions, a cloud camera platform, and MQTT-based device validation.
Designed reliability and reporting architecture for NVR and camera ecosystems across firmware and cloud services.
Developed synthetic-data and edge-inference workflows with Qwen-Image, CLIP, CNNs, and Ambarella CV72S hardware.

Jan 2022 – Mar 2025 Auckland, New Zealand

Fisher & Paykel Appliances · Intermediate Software Engineer

Led development of an automated testing platform integrating hardware, firmware, and higher-level software.
Developed embedded C software and core control algorithms for refrigerator systems.
Maintained Docker, Make, Azure DevOps, and Jenkins CI/CD workflows.

Mar 2021 – Dec 2021 Auckland, New Zealand

The University of Auckland · Teaching Assistant

Supported courses in advanced databases, object-oriented software development, functional programming, and distributed services.

Dec 2019 – Mar 2020 Shenzhen, China

Intellifusion · Machine Learning Intern

Developed and optimized face-recognition models for security and surveillance products.

Publications

Publications page

NeurIPS 2026 Main Track

Population-Aligned Persona Generation for LLM-based Social Simulation

Zhengyu Hu, Jianxun Lian, Zheyuan Xiao, Max Xiong, Yuxuan Lei, and 5 more authors

Advances in Neural Information Processing Systems 39, 2026. Forthcoming.

Builds representative persona sets by aligning LLM-generated profiles with real population-level psychometric distributions.

PDF arXiv BibTeX 38 citations

@inproceedings{hu2025persona,
  title = {Population-Aligned Persona Generation for LLM-based Social Simulation},
  author = {Hu, Zhengyu and Lian, Jianxun and Xiao, Zheyuan and Xiong, Max and Lei, Yuxuan and Wang, Tianfu and Ding, Kaize and Xiao, Ziang and Yuan, Nicholas Jing and Xie, Xing},
  booktitle = {Advances in Neural Information Processing Systems 39},
  year = {2026},
  note = {Forthcoming.},
}

EMNLP 2026 Main Conference

When the Strongest Teacher Is Not the Best Teacher: Student-Centric Answer Selection

Zhengyu Hu^*, Zheyuan Xiao^*, Linxin Song, Fengqing Jiang, Yuetai Li, and 8 more authors

Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, 2026. * Equal contribution.

Selects verified teacher answers by student-specific learning cost instead of assuming the strongest teacher always teaches best.

PDF arXiv Proceedings BibTeX 1 citation

@inproceedings{hu2026studentcentric,
  title = {When the Strongest Teacher Is Not the Best Teacher: Student-Centric Answer Selection},
  author = {Hu, Zhengyu and Xiao, Zheyuan and Song, Linxin and Jiang, Fengqing and Li, Yuetai and Xiong, Zhihan and Lin, Junhao and Liu, Yue and Su, Yao and Hu, Lijie and Ding, Kaize and Xiao, Teng and Poovendran, Radha},
  booktitle = {Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing},
  pages = {31221--31256},
  year = {2026},
  note = {* Equal contribution.},
}

EMNLP 2025 Findings

Explaining Length Bias in LLM-Based Preference Evaluations

Zhengyu Hu, Linxin Song, Jieyu Zhang, Zheyuan Xiao, Tianfu Wang, and 5 more authors

Findings of the Association for Computational Linguistics: EMNLP 2025, 2025.

Explains length bias through desirability and information mass, then introduces a length-aligned evaluation protocol.

PDF arXiv Proceedings BibTeX 79 citations

@inproceedings{hu2025length,
  title = {Explaining Length Bias in LLM-Based Preference Evaluations},
  author = {Hu, Zhengyu and Song, Linxin and Zhang, Jieyu and Xiao, Zheyuan and Wang, Tianfu and Chen, Zhengyu and Yuan, Nicholas Jing and Lian, Jianxun and Ding, Kaize and Xiong, Hui},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2025},
  pages = {6763--6794},
  year = {2025},
  doi = {10.18653/v1/2025.findings-emnlp.358},
}

NeurIPS 2025 Main Conference
Unveiling the Learning Mind of Language Models: A Cognitive Framework and Empirical Study

Zhengyu Hu, Jianxun Lian, Zheyuan Xiao, Seraphina Zhang, Tianfu Wang, and 3 more authors

Advances in Neural Information Processing Systems 38, 2025.

Introduces a cognitive framework for testing how language models learn from instructors, concepts, and experience.

PDF arXiv Proceedings OpenReview Poster BibTeX 11 citations
@inproceedings{hu2025learning, title = {Unveiling the Learning Mind of Language Models: A Cognitive Framework and Empirical Study}, author = {Hu, Zhengyu and Lian, Jianxun and Xiao, Zheyuan and Zhang, Seraphina and Wang, Tianfu and Yuan, Nicholas Jing and Xie, Xing and Xiong, Hui}, booktitle = {Advances in Neural Information Processing Systems 38}, year = {2025}, }

Honors and Awards

Jul 2020

First in Course Award in Computer Graphics and Image Processing

The University of Auckland. Awarded for the top course performance in semester one of 2020.

Skills

Research

LLM evaluation, post-training, LLM agents, social simulation, computer vision, ML systems

Engineering

Python, TypeScript, C, GitHub Actions, Docker, Playwright, AutoGen