I am a Ph.D. candidate in ANSR Lab, supervised by Chair Professor John C.S. Lui (ACM/IEEE Fellow, Fellow of HKAES) at the Department of Computer Science & Engineering, The Chinese University of Hong Kong (CUHK). Before that, I obtained my honor bachelor degree (Advanced Class, GPA: 3.95/4.0) from the Department of Electronics and Information Engineering at Huazhong University of Science and Technology (HUST).

Research

My research resides at the intersection of Agentic Intelligence and Decisional Reinforcement Learning. I focus on building self-evolving LLM systems capable of autonomous reasoning and optimal decision-making in open-ended environments. My core expertise lies in bridging Online Optimization (Bandits) with foundation model alignment, specifically developing inference-time scaling and alignment strategies that bypass the need for expensive global retraining. My goal is to develop AI agents that are not only logically robust but also system-efficient and deeply aligned with complex human intent.

My work focuses on three primary application thrusts:

• Adaptive Reasoning & Inference-time Alignment: Developing sample-efficient online RL (e.g., Bandit-based path selection) to capture real-time feedback, enabling LLMs to self-correct reasoning traces and align with user preferences during inference.

• Autonomous Agent Orchestration: Designing intelligent decisional routing and collaborative strategies (e.g., MoE expert merging, semantic-aware orchestration) to enhance the multi-step problem-solving capabilities of large-scale agent networks.

• Efficient Scaling Laws for Agents: Leveraging learning-based resource orchestration to optimize the trade-off between reasoning compute (Inference-time Compute) and performance, ensuring reliable deployment of agentic workflows in resource-constrained environments.

I always welcome discussions and collaborations. Feel free to reach out via email: “xiangxdai0 [at] gmail.com” (recommended) or “xxdai23 [at] cse.cuhk.edu.hk”.

News

• [02, 2026] One paper accepted by AIJ!
• [01, 2026] One paper accepted by IEEE/ACM TON. One paper accepted by WWW 2026!
• [12, 2025] Four papers accepted by INFOCOM 2026!
• [12, 2025] Granted by NSFC Basic Research Scheme for Ph.D. Student!
• [11, 2025] Two papers accepted by AAAI 2026. One paper accepted by ICDE 2026!
• [06, 2025] Invited as a Session Chair for ACM SIGKDD 2025!
• [05, 2025] Two papers accepted by ACM SIGKDD 2025; also recognized as Outstanding Reviewer (Top 10%) for KDD. One paper accepted by ICML 2025!
• [05, 2025] Our work selected as Best Paper Finalists (Top 5) at ACM SIGMETRICS 2025!
• [05, 2025] Started a visiting internship with Dr. Wei Chen and Dr. Siwei Wang at Microsoft Research Asia (MSRA), Beijing.