Hi! I am Guohui Guan; you can just call me Hui (辉). I am a third-year Ph.D. candidate at the University of Minnesota, Twin Cities, advised by Prof. Chang Ge.
I am interested in the intersection of large language models (LLMs) and data management. I build data systems that use LLMs to integrate, clean, and synthesize data, with guarantees on quality, cost, and privacy.
I am looking for internships for Summer 2027.
I am always happy to chat about research, collaborations, or anything else. Feel free to email me.
News
- Nov 2026NEWOur experimental study on applying data-system execution techniques to speed up embedding-centric AI pipelines has been accepted to WSDM 2027!
- Sep 2026Our paper HotK has been accepted to ICDE 2027!
- Apr 2026I just passed my Ph.D. preliminary exam! Officially a Ph.D. candidate :)
- Feb 2026Our paper Prism has been accepted to SIGMOD 2026!
- Jun 2025Our paper SemForest has been accepted to the TaDA Workshop at VLDB 2025!
Publications
- Uber-Join: Unifying Diverse Table Joins using LLMs with Precision and Cost Guarantees Guohui Guan, Chang Ge, Yeye He Paper
- Faster: Open-World Functionally Augmented Semantic Type Discovery with Foundation Models Guohui Guan, Sachin Konan, Larry Rudolph, Chang Ge Under review, VLDB, 2027
- Understanding Embedding-Centric AI Pipelines: An Experimental Study of Similarity, Reuse, and Ranking Yichen Liu, Guohui Guan, Chang Ge ACM WSDM, 2027 (to appear)
- HotK: Secure Top-K Query Answering over Horizontal Data Federations Junzhou Su, John Carruth, Guohui Guan, Chang Ge IEEE ICDE, 2027 (to appear)
- Prism: Private Relational Data Synthesis with Language Models Guohui Guan, Chang Ge ACM SIGMOD, 2026 · Paper
- SemForest: Semantic-Aware Ontology Generation with Foundation Models Guohui Guan, Sachin Konan, Larry Rudolph, Chang Ge TaDA Workshop at VLDB, 2025 · Paper
- Audio-Visual Cross-Attention Network for Robotic Speaker Tracking Xinyuan Qian, Zhengdong Wang, Jiadong Wang, Guohui Guan, Haizhou Li IEEE/ACM TASLP, 2023 · Paper
- Deep Audio-Visual Beamforming for Speaker Localization Xinyuan Qian, Qiquan Zhang, Guohui Guan, Wei Xue IEEE SPL, 2022 · Paper
Research Experience
-
University of Minnesota, Twin CitiesSep 2024 – PresentGraduate Research Assistant, advised by Prof. Chang Ge
- Designed Uber-Join to unify diverse table joins using LLMs with precision and cost guarantees, handling differences in spelling, format, and meaning.
- Built Faster for open-world semantic type discovery without a predefined label set, equipping each type with executable logic to canonicalize and validate its values.
- Developed Prism for differentially private relational data synthesis with LLMs, privatizing teacher-to-student knowledge transfer via output aggregation with trie-based token pruning and data-dependent noise scaling.
Teaching Experience
-
University of Minnesota, Twin CitiesSpring 2025 – PresentTeaching Assistant
- CSCI 5708: Architecture and Implementation of DBMSSpring 2026
- CSCI 4707: Practice of Database SystemsFall 2025, Spring 2025
Education
-
University of Minnesota, Twin CitiesSep 2024 – May 2029 (expected)Ph.D. in Computer Science Advisor: Prof. Chang Ge
-
University of California, BerkeleySep 2022 – Dec 2023M.Eng. in Electrical Engineering and Computer Sciences Worked with Prof. Gerald Friedland and Eleanor Cawthon
-
National University of SingaporeJan 2020 – Dec 2020Exchange Student in Electrical and Computer Engineering Worked at the HLT Lab with Prof. Haizhou Li and Prof. Xinyuan Qian
-
Chongqing UniversitySep 2017 – May 2021B.Eng. in Electrical Engineering and Automation
Honors
- Departmental GAGE Fellowship2024University of Minnesota, Twin Cities
- EECS Fung Excellence Scholarship2023University of California, Berkeley
- First-Class Scholarship for Merit Students (four times)2018 – 2019Chongqing University
- National Scholarship2018Ministry of Education of China
Outside of Research
I enjoy snowboarding, film photography, and tracking my life through data. I love trying new things and having new experiences.