About
Hi there!🙂 I’m Xunzhuo Zhang (Noah), a game developer and researcher interested in how artificial intelligence can shape games and virtual worlds. I completed my M.A. in Media and Information at Michigan State University.
My current research focuses on LLM-based game agents and AI-driven playtesting. In my master’s thesis, I examined whether LLM agents designed to represent different player types actually produce behavior that aligns with human players, and what this means for the reliable and explainable use of AI in game development. I have the honor of being advised by Dr. Amanda Cote, the Chair of my thesis committee, with Dr. Rabindra Ratan and Dr. Ruth Shillair also as my committee members! My broader interests include game AI, human-AI interaction, procedural content generation, and virtual worlds.
I also have experience in game development, game design, and VR research at MSU’s SPARTIE Lab. My background in computer science, HCI, and media studies helps me approach interactive systems from both technical and human-centered perspectives.
My long-term goal is to help build the next generation of interactive entertainment through more capable game AI and richer virtual worlds. Just like my favorite saying goes: “Be the change!"
Research
Interests
Game AI, LLM-based Agents, Automated Playtesting, Player Modeling, Procedural Content Generation, Human-Centered AI, Player Experience
Paper in Preparation
“LLM-Based Player-Style Agent Generation for Automated Game Playtesting” — master’s thesis, successfully defended July 12, 2026; completed and accepted by Michigan State University. First-author paper in preparation for submission to IEEE Transactions on Games (Aug 2026 – Present).
Thesis Committee: Dr. Amanda Cote (Chair), Dr. Ruth Shillair, Dr. Rabindra Ratan.
- Designed and evaluated style-conditioned LLM agents across Super Mario and Slay the Spire for automated game playtesting.
- Analyzed behavioral traces using feature extraction, PCA, clustering metrics, and human-anchored comparison.
- Examined the gap between prompt-induced behavioral separation and correspondence with real human play, motivating human-anchored validation of AI playtesting agents.
- Refactoring the Slay the Spire evaluation pipeline to remove style-specific rule leakage and preserve LLM decision ownership.
- Standardized player-style prompts, expanded LLM control over non-combat decisions, and implemented reproducible trace logging.
Research Experience
Graduate Research Assistant — Dr. Amanda Cote (Summer 2026)
Michigan State University
- Contributed to an early-stage research project (not yet publicly shared) on an early serious game company.
- Identified and compiled information on company employees involved in the relevant games into spreadsheet to support Dr. Cote’s subsequent research.
Graduate Research Assistant — Dr. Rabindra Ratan (Jun 2026 – Present)
SPARTIE Lab, Michigan State University
- Contributed to an early-stage research project on LLMs in education (not yet publicly shared).
- Contributed to collecting and categorizing relevant papers while coordinating and supervising undergraduate research assistants.
Graduate Research Assistant (Sep 2024 – Dec 2024)
SPARTIE Lab, Michigan State University
- Helped conduct a study on virtual co-working environments and social presence in digital spaces.
- Designed and conducted user studies; performed literature reviews; contributed to internal reports and paper drafts.
Independent Research Project — “Truth Behind the Fake” (Jan 2025 – Aug 2025)
- Designed and developed a serious game addressing misinformation in online communities.
- Conducted user studies, implemented game systems, and maintained organized design documentation.
Posters & Presentations
- Zhang, X. “LLM-Based Player-Style Agent Generation for Automated Game Playtesting.” Peer-reviewed poster accepted for presentation and publication in the Meaningful Play 2026 conference proceedings.
- Zhang, X. “Truth Behind the Fake.” Individual presentation, Play Make Learn 2025.
Selected Work
Purrcel 2D supernatural narrative exploration game (team made). Submitted to the IndieCade Festival; currently under review. I was lead programmer and designer.
Eternal Winter 3D action-adventure game (team made). I implemented gameplay systems (AI, companion, UI, skill tree) as programmer.
A serious game addressing misinformation in online communities. Presented at Play Make Learn 2025.
Procedural map logic and advanced gameplay mechanics in Unity 2D environments.
Background 👉(Download full CV)
Michigan State University, East Lansing, MI, USA (Aug 2024 – Aug 2026)
M.A. in Media and Information (Game Design Track) — GPA: 4.0 / 4.0
City University of Hong Kong, Hong Kong SAR, China (Sep 2020 – Jun 2024)
B.Sc. in Computer Science