It is reported that Yu Bowen, a core member of the original Qwen team, has joined ByteDance's Seed team

According to reports, after Lin Junyang, the head of large model technology at Alibaba's Tongyi Lab, left, Yu Bowen, the former head of post-training for Qwen (Qianwen), was recently reported to have officially joined ByteDance as the head of post-training for the Seed team's visual model and multimodal interaction team. Although ByteDance has not yet made an official response, multiple sources indicate that this personnel change follows the organizational restructuring at Alibaba's Tongyi Lab in early March.

This restructuring split the previously vertically integrated Qwen team into parallel modules such as pre-training, post-training, and multimodal processing, leading to the departure of several key technical personnel, including Yu Bowen, Kaixin Li (core contributor to Qwen3.5/VL), and Hui Bingyuan, who had previously joined Meta. Yu Bowen, a technical expert from "Alibaba Star," led the development of the Qwen series of chat models and has deep expertise in areas such as multimodal alignment.

The Seed team at ByteDance, which was taken over by Dr. Wu Yonghui, former vice president of research at Google DeepMind, in February 2025, reports directly to CEO Liang Rubo. The department has already developed the Seed2.0 series of base models and the Seedance2.0 video generation model. As global competition in large models enters a critical phase focused on multimodal and visual understanding, ByteDance's recruitment of core talent from Tongyi reflects the high importance placed by leading companies on post-training technology, and also signals that talent mobility in China's large model industry is accelerating the restructuring of the underlying technology landscape.