Profile

My research interests focus on multimodal learning and data-centric artificial intelligence. Currently, I am working on ultra-large-scale, efficient vision-language pre-training (at the level of 1000 GPUs and 10 billion samples) and multimodal large-scale models. I obtained my Bachelor's and Master's degrees from Sun Yat-sen University (SYSU). I completed my PhD at the National University of Singapore (NUS) in three years.

My GitHub has over 5,000 stars, and I have years of experience training models with approximately 1,000+ GPUs. For details, see: https://scholar.google.com/citations?user=UtaAVacAAAAJ&hl=zh-CN.

I have published 13 first-author or co-first-author papers in CCF-A conferences or JCR Q1 journals. For details, see: https://scholar.google.com/citations?user=UtaAVacAAAAJ&hl=zh-CN.

Personal homepage: http://fingerrec.github.io.

Key Information:

Our research group has access to a maximum of 1,000 GPUs (mainly V100, A100, and Huawei 310A) for training. 

Through my direct introductions, three students from our group are currently interning or have completed internships at Microsoft (no interview required). 

Team members come from institutions such as the National University of Singapore, Peng Cheng Laboratory, Tsinghua Shenzhen International Graduate School, Chinese Academy of Sciences, Zhejiang University, and the University of Science and Technology of China.

Research Directions:

1. Unified multimodal generation & understanding models.

2. Post-training for mainstream multimodal models like GPT-4o, DeepSeek-VL, and GeminiPro to address capability gaps.

3. Large-scale pre-training.

4. Reasoning & Chain-of-Thought in Large Vision Models (LVM).

Internships & Further Studies:

- Internship referrals: Microsoft, Sea AI Lab, Alibaba, Tencent ARC, Tencent Youtu Lab, ByteDance.

- Academic recommendations: Singapore, UK, USA, Hong Kong, etc.

Recruitment (Research Assistants/Master’s Students/Undergraduate Research Assistants):

I have a calm personality and dislike conflict. Joining the group requires an interview; only reliable and motivated individuals should contact me—please no casual inquiries. We focus on coding-based research, with preference given to students with strong competition/coding skills. I am good at organizing students, maintaining lab relationships, and prioritize candidates who are good at basketball.