Hi there! I am Ruotian Wu, a Master’s student at University of Waterloo, supervised by Professor Pascal Poupart.

I earned my Bachelor of Mathematics with a double major in Computer Science and Statistics from the University of Waterloo. My current research focuses on Large Language Models (LLMs), particularly alignment and decoding algorithms. Additionally, I am collaborating with Scribendi on AI-driven document editing. I am also exploring ways to enhance LLM agents to improve their capabilities on different game settings. I am always open to new ideas, collaborations, and industry partnerships. If you’re interested in working together, feel free to reach out!

🔥 News

• 2025.02:  🎉🎉 Our latest work FaRMA is now available on ArXiv and is currently under review for ICML 2025. This research introduces a novel reward model architecture to reduce the computational overhead of traditional Reward Guided Text Generation (RGTG) methods. We also propose a new training paradigm that is more principled and efficient.
• 2025.02:  🎉🎉 Our work on a new search algorithm is available on Arxiv and is currently under review for ICML2025. This work proposes an uncertainty-guided tree search algorithm for settings where the reward function is a log-likelihood function of the paths.
• 2024.10:  🎉🎉 Our research on enhancing RGTG is now available on ArXiv. This work provides a comprehensive analysis of existing RGTG methods and improves the performance by leveraging partial-sequence preference data to train more effective reward models.

📝 Publications

• Towards Cost-Effective Reward Guided Text Generation, Ahmad Rashid ^*, Ruotian Wu ^*, Rongqi Fan, Hongliang Li, Agustinus Kristiadi, Pascal Poupart Arxiv
• Uncertainty-Guided Likelihood-Tree Search, Julia Grosse, Ruotian Wu, Ahmad Rashid, Philipp Hennig, Pascal Poupart, Agustinus Kristiadi Arxiv
• A Critical Look At Tokenwise Reward-Guided Text Generation, Ahmad Rashid, Ruotian Wu, Julia Grosse, Agustinus Kristiadi, Pascal Poupart Arxiv
• Achieving fairness in team-based FPS games: A skill-based matchmaking solution, Ruotian Wu, Xiangcheng Meng, Haoshen Chen, Zixuan Zhu, Bo Wang MLA 2023

🎖 Honors and Awards

• 2024.09 International Master’s Award of Excellence, University of Waterloo
• 2024.04 Graduation with Dean’s Honors, University of Waterloo
• 2019.09 President’s Scholarship, University of Waterloo

📖 Educations

• 2019.09 - 2024.04, Bachelor of Mathematics, Honours Computer Science, University of Waterloo
• 2019.09 - 2024.04, Bachelor of Mathematics, Honours Statistics, University of Waterloo
• 2024.09 - 2026.09(expected), Master of Mathematics, Computer Science, University of Waterloo

🧑‍🏫 Teaching

• 2024.10 - 2025.03, Teaching Assistant, Green-AI bootcamp, vector Institute
• 2024.09 - 2024.12, Teaching Assistant, CS246 Object Oriented Programming
• 2021.05 - current, Tutor, Cambridge IGCSE and A-level Computer Science (0478 & 9618)

💬 Invited Talks

• 2024.12, Presented the latest advancements in LLM alignment at the Green-AI Bootcamp, organized by the Vector Institute.

💻 Internships

• 2023.09 - 2024.04, Vector Institute, Research Assistant
• 2023.01 - 2023.04, Manulife/JohnHancock, Platform Reliability Engineer
• 2022.05 - 2022.08, HUAWEI, Support Engineer
• 2021.05 - 2021.12, Youth STEM Academy, Computer Science Tutor
• 2020.08 - 2020.12, Hande-China, Management Consultant