Hi, I’m Wenjia Jiang, currently a research assistant at AGI Lab, Westlake University, advised by Prof. Chi Zhang. I received my Bachelor’s degree in software engineering from the School of Software, Henan University of China.

I am applying to Ph.D programs this year (26 fall). My current research interests lie in large language models (LLMs), multimodal learning, and artificial intelligence. I am particularly interested in developing LLM-based agents. Download my CV

Selected Research Projects

AppAgentX: Evolving GUI Agents as Proficient Smartphone Users

A novel evolutionary framework for GUI agents that enhances operational efficiency through action evolution.

Project Page | PDF | Code

Learning to Be A Doctor: Searching for Effective Medical Agent Architectures

Accepted by ACM MM 2025
An innovative approach to searching AI agent architectures capable of medical reasoning and diagnosis in medical scenarios.

PDF

Current Research

ACGs: Agents Can Generate SQL by Querying Trees Like Expert

Supervised jointly by Prof. Chi Zhang and Prof. Yiwei Wang (University of California, Merced). Designed a novel framework to enhance the accuracy of LLM-based agents in complex NL2SQL tasks for enterprise scenarios. Simulated human data scientists’ database exploration via Monte Carlo search, leveraging the discovered structural information to guide SQL generation for challenging queries. Preliminary evaluation on benchmarks such as Spider 2.0 shows promising results. (📝 Under review at ICLR 2026)

Paper2Beamer

Developed a system leveraging an LLM-based agent to convert academic papers into presentation slides. Investigated interactive slide modification with humans in the loop, drawing on concepts from human-computer interaction and educational theory. (📝 Under review at CHI 2026)

Skills

• Language: English (IELTS 6.5)

• LLM Applications: Skilled in developing applications with LangChain and Camel AI, prompt engineering, integrating OpenAI API, and building Retrieval-Augmented Generation (RAG) systems.

• Data & Storage: Experienced with vector databases, graph databases, and data formats such as JSON and HDF5 for efficient data management and retrieval.

• Deep Learning: Proficient in PyTorch, Transformer architectures, and Computer Vision tasks; hands-on experience with model training, fine-tuning, and deployment.

• Tools & DevOps: Proficient with Git, Docker, FastAPI, Jupyter, and Hugging Face ecosystem for collaborative development, deployment, and experimentation.

• Professional Software: LaTeX, SPSS.