Hi, my name is Xutong (James) Liu. I am now a Tenure-Track Assistant Professor at the University of Washington, affiliated with the Department of Computer Science & Systems at the Tacoma School of Engineering & Technology. At UW, I am building the Learning, Evaluation, and Advanced Decision-making (LEAD) research lab and actively recruiting Ph.D. students, master students, and undergraduate interns in Fall 2026, check group information for details if you are interested in joining us.

Previously, I was a postdoctoral researcher in LIONS research group at Carnegie Mellon University, advised by Prof. Carlee Joe-Wong. Before that, I was a visiting postdoc in SOLAR Lab at the University of Massachusetts Amherst, advised by Prof. Mohammad Hajiesmaili, and a postdoctoral fellow in ANSR Lab at the Chinese University of Hong Kong (CUHK), advised by Prof. John C.S. Lui (IEEE/ACM Fellow). I received my Ph.D. degree from the Computer Science and Engineering Department at CUHK in 2022, proudly advised by Prof. John C.S. Lui. Prior to that, I received my bachelor’s degree with an honored rank (top 5%) from the University of Science and Technology of China (USTC) in 2017.

My Research

My research focuses on developing structure-aware online learning and reinforcement learning (RL) algorithms that leverage inherent action, feedback, and agent structures—such as smoothness, sparsity, and clustering—to enable data-efficient, scalable, and robust decision-making in networked systems with fused offline and online data.

I aim to bridge theory and real-world applications across LLM systems, edge/cloud computing, and conversational recommender systems, guided by three core questions:

1. Data Efficiency: How much offline and/or online data is needed to achieve near-optimal policies?
2. Scalability: How can we design algorithms that scale efficiently across large action spaces and multiple agents?
3. Robustness: How can learning remain stable and adaptive in dynamic, uncertain, and heterogeneous environments?

My recent work focuses on building:

🚀 Structure-aware Online Learning/Reinforcement Learning (RL):

• Hybrid Reinforcement Learning with Fused Offline-Online and Relative-Absolute Data:
  [NeurIPS ‘25], [ICML ‘25b], [KDD ‘25], [arXiv ‘25]

• Reinforcement Learning with Large Action Spaces:
  [ICML ‘25a], [ICML ‘24]
• Scalable Combinatorial Decision-making under Uncertainty:
  [NeurIPS ‘22], [ICML ‘21, 🎤️Long Oral]
• Generalizable Combinatorial Online Learning with Function Approximation:
  [SIGMETRICS ‘25, 🏆Best Paper Runner-Up], [INFOCOM ‘24], [ICML ‘23]
• Robust Multi-agent Online Learning in Heterogeneous and Unreliable Environments:
  [ICLR ‘25], [SIGMETRICS ‘25], [INFOCOM ‘25], [AAAI ‘24], [NeurIPS ‘23], [UAI ‘23], [AISTATS ‘23], [ICLR ‘23], [UAI ‘23]

🛜 Decision-making for Network Applications:

• Cost-effective Large Language Model (LLM) Training/Serving Systems:
  [arXiv ‘25a], [arXiv ‘25b], [arXiv ‘25c], [arXiv ‘24]
• Edge/Cloud Computing, Multimedia Networking, and IoT Systems:
  [IEEE/ACM TON], [ACM MM ‘24], [INFOCOM ‘23], [INFOCOM ‘18], [IEEE TMC]
• Conversational Recommendation Systems and Social Network Marketing:
  [KDD ‘25], [AAAI ‘23], [AISTATS ‘22], [IEEE TKDE]

Selected Recent Publications

• [Preprint] HiLoRA: Adaptive Hierarchical LoRA Routing for Training-Free Domain Generalization
  Ziyi Han, Huanyu Wang, Zeyu Zhang, Xiangxiang Dai, Xutong Liu, John C.S. Lui.
  [arXiv]

• [Preprint] Faster, Smaller, and Smarter: Task-Aware Expert Merging for Online MoE Inference
  Ziyi Han, Xutong Liu, Ruiting Zhou, Xiangxiang Dai, John C.S. Lui.
  [arXiv]

• [Preprint] Offline Clustering of Preference Learning with Active-data Augmentation
  Jingyuan Liu, Fatemeh Ghaffari, Xuchuang Wang, Xutong Liu#, Mohammad Hajiesmaili, Carlee Joe-Wong.
  [arXiv]

• [NeurIPS '25] Learning Across the Gap: Hybrid Multi-armed Bandits with Heterogeneous Offline and Online Data
  Qijia He, Minghan Wang, Xutong Liu, Zhiyong Wang, Fang Kong.
  The Thirty-nineth Conference on Neural Information Processing Systems (NeurIPS), 2025.
  [Openreview]

• [SIGMETRICS ‘25, 🏆Best Paper Runner-Up] Combinatorial Logistic Bandits
  Xutong Liu, Xiangxiang Dai, Xuchuang Wang, Mohammad Hajiesmaili, John C.S. Lui.
  The ACM International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2025.
  Best Paper Runner-Up at SIGMETRICS 2025.
  [arXiv] [code] [slides]

• [NeurIPS ‘22] Batch-Size Independent Regret Bounds for Combinatorial Semi-Bandits with Probabilistically Triggered Arms or Independent Arms
  Xutong Liu, Jinhang Zuo, Siwei Wang, Carlee Joe-Wong, John C.S. Lui, Wei Chen.
  The 36th Conference on Neural Information Processing Systems (NeurIPS), 2022.
  [arXiv] [paper] [slides] [poster]

• [UAI ‘22] Federated Online Clustering of Bandits
  Xutong Liu, Haoru Zhao, Tong Yu, Shuai Li, John C.S. Lui.
  The 38th Conference on Uncertainty in Artificial Intelligence (UAI), 2022.
  [paper] [arXiv] [slides] [poster] [code]

• [ICML ‘21, 🎤️Long Oral] Multi-layered Network Exploration via Random Walks: From Offline Optimization to Online Learning
  Xutong Liu, Jinhang Zuo, Xiaowei Chen, Wei Chen, John C.S. Lui.
  The 38th International Conference on Machine Learning (ICML), Long Oral, 2021.
  [paper] [arXiv] [slides] [poster] [video]

News

• Sept. 2025: Our work on Hybrid Multi-armed Bandits with Heterogeneous Offline and Online Data is accepted by NeurIPS 2025.
• Sept. 2025: I am joining the University of Washington - Tacoma as a Tenure-Track Assistant Professor In CSS.
• June. 2025: I am invited as a TPC for ACM SIGMETRICS 2026 and ACM e-Energy 2026!
• May. 2025: Our paper “Combinatorial Logistic Bandits” has been selected as one of the Best Paper Runner-Up at SIGMETRICS 2025!
• Dec. 2024: We are excited to co-organize the 3rd Annual Workshop on Learning-Augmented Algorithms: Theory and Applications at ACM SIGMETRICS 2025. The workshop will take place at Stony Brook University, New York, USA. For more details, visit the official workshop website.
• Dec. 2024: Our works on (1) robust combinatorial contextual bandits and (2) online learning algorithms to learn the best quantum path have been accepted by INFOCOM 2025.
• Oct. 2024: Our work on combinatorial bandits with logistic function approximation has been accepted by ACM SIGMETRICS 2025.
• Sept. 2024: I am joining Carnegie Mellon University as a postdoctoral researcher advised by Prof. Carlee Joe-Wong.
• May. 2024: Our works on (1) combinatorial bandits view to solve episodic RL and (2) quantum algorithm for online exp-concave optimization are accepted by ICML 2024.
• Oct. 2023: I am visiting the University of Massachusetts Amherst as a visiting scholar advised by Prof. Mohammad Hajiesmaili.
• Sept. 2023: Our work on online clustering of bandits with misspecified user model is accepted by NeurIPS 2023.
• April. 2023: Our work on contextual combinatorial bandits with probabilistically triggered arms is accepted by ICML 2023.
• April. 2023: I was awarded RGC Postdoctoral Fellowship (one of 50 awardees globally)!