Databases · AI · Systems
Zhengle Wang
Ph.D. Student in Computer Science · Purdue University
I build database systems and AI for data management. At Purdue, I work with Prof. Chunwei Liu on efficient semantic query processing and realistic workloads for cloud analytics.
I am the student lead of JEVDB. My broader interests include execution-aware workload synthesis and infrastructure for agentic systems, from trajectory management to reproducible evaluation.
- Semantic data systems
- Cloud analytics & benchmarking
- Agentic systems infrastructure
Research highlights
Semantic data systems
JEVDB
JEVDB is a next-generation database engine for scalable semantic query processing.
It brings AI-powered understanding into the database, with the vision of making queries over unstructured data as natural and scalable as queries over structured tables.
Cloud analytics benchmarking
Workload Synthesis
Workload Synthesis brings production realism to benchmarking for modern data systems.
Through PBench and ResQ, it turns production traces into runnable workloads, helping us build and evaluate systems for the complexity and diversity of real-world use.
Selected publications
All publications2026
- Prune First, Decide Fast: Scalable Semantic Query Processing with JEVDB
- Fail Loudly: An Auditable Runtime for Agentic Data Analysis
- ResQ: Realistic Performance-Aware Query Generation
2025
- PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking
News
All updates| Oct 1, 2026 | We released JEVDB, a semantic database system for scalable query processing (paper). |
|---|---|
| Aug 24, 2026 | Started Ph.D. in Computer Science at Purdue University, working with Prof. Chunwei Liu on AI for databases and cloud benchmarking. |
| May 1, 2026 | Joined Analogy AI as a Research Scientist Intern, building TraceForge infrastructure for computer-use agent evaluation. |
| Oct 1, 2025 | Internship at Meituan, Hotel and Travel Recommendation Algorithm Team. |
| Jun 1, 2024 | Internship at Poizon (Dewu), Community Search Team, focusing on search ranking and relevance. |
Recent writing
All posts| Jul 30, 2026 | A Database Paradigm for the Future |
|---|---|
| May 23, 2026 | Agentic System Workload Optimization |
Industry experience
Before joining Purdue, I collaborated with researchers and engineers from Renmin University of China, Massachusetts Institute of Technology, and Amazon Web Services on cloud database benchmarking and system optimization.
-
Analogy AI
May 2026 – Aug 2026Research Scientist Intern · Remote · Palo Alto, CA
Built TraceForge infrastructure for evaluating computer-use agents.
-
Meituan
Oct 2025 – Dec 2025Machine Learning Intern · Hotel & Travel Recommendation Team · Beijing, China
Worked on cross-domain and multi-scenario recommendation modeling across hotel, homestay, and ticket-booking services. Explored shared representation learning and scenario-specific adaptation strategies for integrating heterogeneous user behaviors and item features across different travel-related scenarios.
-
Poizon Inc.
Jun 2024 – Mar 2025Machine Learning Intern · Community Search Team · Beijing, China
Worked on machine learning methods for large-scale content search and recommendation, including relevance-aware ranking, personalized pre-ranking, and model distillation for efficient retrieval/ranking pipelines. This experience exposed me to production-scale ML systems and motivated my broader interest in building reliable, data-driven infrastructure for real-world workloads.