Hi!!! I am a 4th-year PhD candidate in CIS at Temple University 🦉, advised by Prof. Eduard Dragut. I received my Master degree in Artificial Intelligence and Automation from Huazhong Univeristy of Sci & Tech (HUST) advised by Prof. Min Ouyang. I received my bachelor’s degree in Electrical Engineering and Automation from Wuhan University of Technology.

I explore Agentic AI for Education, Social Good, and Scientific Research, with my work focusing on two main directions:

• Agentic AI and Distillation: Proposing novel Agentic AI methods to address challenges in social content analysis and GUI navigation agents, and distilling agentic reasoning into smaller LLMs through fine-tuning or reinforcement learning (RL).

• Human-AI Collaboration: Advancing Agentic AI with human-in-the-loop for qualitative analysis, meme understanding, and long scientific document analysis, pioneering active and efficient self-improvement from human feedback.

🌈 Updates

• Apr 17, 2026 - Joined the AIX Summit in New York.
• Apr 6, 2026 - CentaurTA has been accepted as Findings of ACL 2026.
• Mar 26, 2026 - Accepted Research Summer Intern Offer from Samsung.
• Feb 20, 2026 - Served as a reviewer for KDD 2026.
• Dec 10, 2025 - Passed the Prelim II exam and advanced to PhD candidacy.
• Dec 2, 2025 - Joined NeurIPS 2025 in San Diego.
• Nov 10, 2025 - Joined CLunch meeting hosted by Penn NLP.
• Nov 8, 2025 - One paper has been accepted to the AAAI 2026 Main Track.

🪶 Selected Publication

CentaurTA: A Self-Improving Human-Agents Collaboration Framework for Thematic Analysis
Lei Wang, Min Huang, Eduard Dragut
Findings of ACL 2026, appearing.

A Multi-Task Multi-Agent Framework for Meme Understanding
first-author paper
(Under Review)

DanceHA: A Multi-Agent Framework for Document-Level Aspect-Based Sentiment Analysis
Lei Wang, Min Huang, Eduard Dragut
AAAI 2026 Main Track (acceptance rate 17.6%), appearing.
Code and Dataset

The Overlooked Repetitive Lengthening Form in Sentiment Analysis
Lei Wang, Eduard Dragut
EMNLP 2024 Findings
Code and Dataset

🚀 Selected Projects & Demos

Autism-Alexa: We propose Autism-Whisper, a customized speech recognition model tailored for individuals with autism. We further seamlessly integrates Autism-Whisper with Amazon Alexa, paving the way for autistic individuals to enjoy the convinent of advanced smart home technologies. Watch our demo.

GUI-Crawler: We design an efficient vision-based GUI-Agent that overcomes SOTA anti-crawling defenses, enabling scalable data collection from complex websites like LinkedIn and X.com. Watch our demo.