Xiaojun Wu

About Me

I’m a full-time PhD candidate at HKUST Guangzhou, and I’m also an affiliate researcher at the IDEA Research, working on Large Language Models. I am one of the earliest developers in China to conduct research on Pretrain Models, Vision-Language Model, and Diffusion-based AIGC model. Internship and work experience include internet companies, state-owned enterprises, and research institutes. Rich in technical passion and curiosity, enjoys sharing and exploring Data and Knowledge, Multimodal, Generative AI, and General Artificial Intelligence. My current research focuses on LLM model training, Harness Engineering, datasets and data synthesis, and benchmark development, with an emphasis on Mathematical Modeling tasks.

News

• [Jun. 2026] 🎉 Honored to be a co-author and contributor of SkillsBench, contributing tasks and BenchFlow SDK development.
• [Apr. 2026] 🎉🎉 2 papers about LLM Data Engineering and Data Synthetic are accepted to ACL 2026 (1 Findings paper and 1 System Demostration paper).
• [Nov. 2025] 🎉 Our paper about Anomaly Detection on Dynamic Graphs is accepted to AAAI 2026.
• [Sept. 2025] 🎉 Our paper about Financial Large Language Models is accepted to EMNLP 2025 Findings.
• [Jan. 2024] 🎉 Taiyi-Diffusion-XL Technical Report is released at Arxiv and HuggingFace
• [Sept. 2023] 🎉 Ziya-VL Technical Report is released at Arxiv and HuggingFace and is reported by HuggingFace.
• [Nov. 2022] 🎉 Our GTS-Engine is reported by JIQIZHIXIN·机器之心.
• [Oct. 2022] 🎉 Our Fengshenbang-LM is reported by JIQIZHIXIN·机器之心.
• [Oct. 2022] 🎉 Our paper about Multimodal Dialogue is accepted to International Journal of Machine Learning and Cybernetics.
• [Oct. 2022] 🎉 Our paper about Noise Learning is accepted to COLING 2022.

Experiences

Education

• [2024.09 - Now] Joint PHD program of HKUST(GZ) and IDEA Research, co-supervised by Prof.Jia LI and Prof.Jian GUO
• [2018.09 - 2020.07] Master’s in Information Science, supervised by Prof.Jimin Wang, SCIE Lab, Peking University
• [2014.09 - 2018.07] Bachelor’s in Information Management and Information Systems, Xidian University

Intern and Work

• [2021.07 - Now] from NLP Algorithm Engineer to affiliate LLM Researcher, IDEA, Shenzhen
• [2020.07 - 2021.07] Algorithm Engineer, The Smart City Research Institute of CETC, Shenzhen
• [2019.07 - 2020.05] Big Data Development Intern, ByteDance, Beijing
• [2018.07 - 2018.10] Data Analysis Intern, Chinese Academy of Science and Technology for Development, Ministry of Science and Technology (MOST), Beijing

Research Interests

• Agent: Multi-Agent, Agent Evolution, Memory and Knowledge, Tool use, Reinforcement Learning
• Multi-modal: Multi-modal LLM, Diffusion model
• NLP: LLM, Natural Language Understanding, Information Retrival, Natural Language Generation
• Application: Mathematical Modeling, Financial AI, Operations Research

Selected Publications

Code Projects

Citation Map

Services

Teaching Assistant

• HKUST (GZ) 2026 Spring: UFUG 2602 - Data Structure and Algorithm Design
• Peking University 2018 Fall: Web Data Mining

© 2025 Xiaojun Wu. All rights reserved. Powered by Jekyll and Minimal Light theme.