Zhenda Xie

Xie helped create techniques that let image models learn from pictures without requiring extensive human labels. This work enabled larger and more capable vision systems, and he later contributed to the training and scaling of DeepSeek’s language models.