Zhensheng Luo Ph.D. Student in Software Engineering, Zhejiang University (Advisor: Prof. Sai Wu). I work on LLM-powered data systems, focusing on the interface between natural language reasoning and reliable database execution. Research Areas: NLP for Data, AIxDB/DBxAI, Agentic LLM Systems, Post-training for Data Tasks. GitHub • Email

• Zhejiang University, Hangzhou, China
  Ph.D. in Software Engineering, Sep 2025 - Present
• Beijing University of Posts and Telecommunications, Beijing, China
  B.Eng. in Software Engineering, Sep 2021 - Jun 2025

• QBridge: Bridging Natural Language and SQL via Gold Query Rewriting with Agentic Refinement
  Zhensheng Luo, Sai Wu, Yuan Qiu, Chang Yao, Gang Chen, Xiu Tang*
  ACL 2026 Main Conference (Poster)

• LogicDB: Co-Optimizing LLM Reasoning and Database Execution for Analytics
  First Author, Under Review (CCF-A)

• Institute of Computing Technology, CAS — Research Intern (Autonomous Driving), Sep 2023 - Jun 2024
  Optimized VectorNet-based trajectory prediction via global pruning and subgraph encoding redesign.

• Data Intelligence Lab, ZJU — Research Intern, Feb 2024 - Aug 2024
  Built data-agent and multimodal QA prototypes for real-world analytical tasks.

• Meituan, Advertising Division — Research Intern, Oct 2024 - Dec 2024
  Built and fine-tuned LLM-based service agents for intent understanding and high-quality domain responses.

• Robust NL-to-Analytics generation with iterative self-refinement.
• Co-design of LLM inference strategies and physical query execution.
• Agentic RL and OPD-style post-training for data-centric reasoning.

_{Open to research collaborations on LLM systems and AI-native data infrastructure.}