Hexiang Hu is a Member of Technical Staff at SpaceXAI, where he works on multi-modal pre-training and native image generation for Grok. Previously he was a Research Scientist at Google DeepMind and a core contributor to Gemini and Imagen 3. He earned his Ph.D. in Computer Science from the University of Southern California (USC), advised by Prof. Fei Sha. His work centers on multi-modal foundation models — pre-training, understanding, and generation. [ CV ]

Hexiang Hu is a Member of Technical Staff at SpaceXAI, where he works on multi-modal pre-training and native image generation for Grok (Aurora, Grok Imagine). Prior to that, he was a Research Scientist at Google DeepMind and a core contributor to Gemini and Imagen 3. He earned his Ph.D. degree in Computer Science from the Viterbi School of Engineering at the University of Southern California (USC), advised by Prof. Fei Sha; he began his Ph.D. at the Henry Samueli School of Engineering and Applied Science at the University of California, Los Angeles (UCLA), before transferring to USC. He earned dual Bachelor’s degrees in Computer Science from Zhejiang University and Simon Fraser University with honors and worked with Prof. Greg Mori. He previously worked at Google Research, Facebook AI Research, Intel Labs, and Amazon AWS AI.

His work centers on multi-modal foundation models — pre-training, understanding, and generation. To him, understanding is ultimately grounded: the meaning of language lives not in the co-occurrence statistics of words, but in the situations and interactions of the physical world it describes — so machines that truly understand must go beyond text, into the multi-modal, embodied world. He holds broader interests in Machine Learning, Natural Language Processing, and Computer Vision. [ CV ]