According to reports, Tian Yonglong, a former researcher at OpenAI, has recently joined Tencent and is expected to serve as the head of the multimodal model direction of Huan Yuan, participating in the development of visual language models. If this personnel change is finalized, it will be another researcher from OpenAI's core research team to join Tencent's Huan Yuan, following Yao Shunyu.

The connection between Tian Yonglong and Yao Shunyu dates back to their time working together at OpenAI. In September 2025, 28-year-old Yao Shunyu left OpenAI to join Tencent as Chief AI Scientist in the "CEO/President Office," directly reporting to President Liu Zhiping. Both were part of the research team at OpenAI, experiencing the rapid iteration of the GPT series and multimodal technologies. Now they are once again gathering under the umbrella of Tencent's Huan Yuan.

Outstanding academic background from Tsinghua University to MIT, with nearly 29,000 Google Scholar citations

Tian Yonglong's academic background represents the typical path of top talent in China's computer vision field. He graduated from Tsinghua University, then obtained a master's degree from the Chinese University of Hong Kong, where he studied under Tang Xiaoe and Wang Xiaogang, laying the foundation for visual recognition and deep learning at MMLab. Later, he entered MIT's EECS department, pursuing his doctorate under the guidance of generation model scholar Phillip Isola, graduating in 2022, shifting his focus from traditional visual recognition to generative models and representation learning.

After obtaining his doctorate, Tian Yonglong worked successively as a senior research scientist at Google Research and Google DeepMind before joining OpenAI as a Member of Technical Staff. His Google Scholar citations have reached 28,778 times, with an h-index of 35, and he had 3,436 citations in 2026 alone. One of his most influential works is supervised contrastive learning, which extended the contrastive learning framework to supervised scenarios, influencing the design ideas of subsequent visual pre-training models such as SimCLR and MoCo.

Huan Yuan's multimodal team continues to recruit talent, accelerating the formation of its talent puzzle

Tian Yonglong's addition is the latest in a series of personnel adjustments made by Tencent's Huan Yuan multimodal team over the past year. At the end of 2024, the former head of the multimodal coordination, Liu Wei, left to start his own business focusing on video generation, and the direction was temporarily managed directly by Tencent's vice president, Jiang Jie. In April 2025, Tencent officially established the Large Language Model Department and the Multimodal Model Department within the TEG system, addressing the previous issues of "virtual teams and scattered resources" from an organizational structure perspective.