Rui Wang (çç)
Computer Science & Artificial Intelligence
I am a third-year undergraduate student at The Hong Kong University of Science and Technology (HKUST), pursuing a Bachelor of Engineering in Computer Science with an Extended Major in Artificial Intelligence. I completed a semester exchange at The University of Texas at Austin (UT Austin) in May 2026. I am currently working at AQUMON (Shenzhen) as an AI Large Language Model Research Intern through the HKUST CSE Co-op Program.
My research interests broadly lie in Large Language Models (LLMs) and their reasoning capabilities. Specifically, I focus on:
- LLM Decision-making
- Uncertainty Quantification and Confidence Calibration
- RAG Robustness
- Rule Induction
Feel free to reach out - my contact information is available in the sidebar.
đĽ News
đ Research
PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems
Jiayu Liu, Qihan Lin, Cheng Qian, , Emre Can Acikgoz, Xiaocheng Yang, Jiateng Liu, Zhenhailong Wang, Xiusi Chen, Heng Ji, Dilek Hakkani-Tur
- Introduced PlanBench-XL, an interactive benchmark with 327 retail tasks and 1,665 tools for evaluating long-horizon planning under retrieval-limited tool visibility
- Showed that even leading models struggle to adapt when tool paths are blocked, highlighting major weaknesses in robust agentic planning for large, imperfect tool ecosystems
Rethinking Prospect Theory for LLMs: Revealing the Instability of Decision-Making under Epistemic Uncertainty
, Qihan Lin, Jiayu Liu, Qing Zong, Tianshi Zheng, Dadi Guo, Haochen Shi, Weiqi Wang, Yangqiu Song
- Designed a three-stage workflow to evaluate LLM decision-making under uncertainty, estimating Prospect Theory parameters and testing whether the framework meaningfully fits model behavior
- Showed that Prospect Theory is not consistently reliable for interpreting LLMs, and that the inferred behavior is especially unstable under epistemic uncertainty expressed through linguistic markers
NOVA: NOise-aware Verbal Confidence CAlibration for Robust Large Language Models in RAG Systems
Jiayu Liu, , Qing Zong, Yumeng Wang, Cheng Qian, Qingcheng Zeng, Tianshi Zheng, Haochen Shi, Dadi Guo, Baixuan Xu, Chunyang Li, Yangqiu Song
- Proposed NOVA, a noise-aware verbal confidence calibration framework for RAG systems that improves reliability when retrieved evidence is contradictory or irrelevant
- Built supervision from NOVA Rules and showed strong improvements in expected calibration error across both in-domain and out-of-domain settings
đŤ Education
The Hong Kong University of Science and Technology
Sep 2023 - May 2027Bachelor of Engineering in Computer Science - Extended Major in AI
The University of Texas at Austin
Jan 2026 - May 2026Semester Exchange, Electronic and Computer Engineering
University College London
Jul 2024Summer School
đź Internships
AQUMON
Jun 2026 - PresentAI Large Language Model Research Intern, Shenzhen
đ¤ Services
Teaching Assistant
HKUSTCOMP1021: Introduction to Computer Science
Peer Mentor
HKUSTCSE Peer Mentor Program
âď¸ Get In Touch
I am always open to discussing research opportunities and collaborations. Please feel free to reach out.