Li-Jia Li

Timeline

1. 2007

   A robot vision challenge

   As a UIUC ECE graduate student, Li joins the Princeton–Illinois team that wins first place in the AAAI Semantic Robot Vision Challenge in 2007.

2. 2007–2009

   ImageNet and scene understanding

   In 2007, she coauthors OPTIMOL, an AAAI framework for online picture collection via incremental model learning. In 2009, she coauthors ImageNet, a large-scale hierarchical image database, and a CVPR model with Richard Socher and Li Fei-Fei that classifies scenes, segments object components, and annotates images with tags.

3. 2010

   Object Bank at Stanford

   In 2010, she and colleagues develop Object Bank, an image representation that encodes semantic and spatial information through generic object detectors and supports scene classification, annotation, and retrieval.

4. 2016

   Visual Genome connects images and language

   In 2016, she coauthors Visual Genome, a dataset that models visual relationships through dense annotations of objects, attributes, and relations within images.

5. 2016–2018

   Research leadership at Google Cloud

   In 2016, she joins Google Cloud AI as head of R&D after leading research at Snap and Yahoo Labs. In November 2018, she leaves Google and plans to apply AI to health care while teaching at Stanford. She holds a PhD in computer science from Stanford.

6. 2019

   A Longuet-Higgins Prize

   In 2019, the IEEE PAMI Longuet-Higgins Prize recognizes ImageNet as the retrospective most impactful paper from CVPR 2009.

7. 2025

   Joining Unity's AI Council

   In 2025, she joins Unity's AI Advisory Council as one of its first two members. At LiveX, she leads development of advanced human-like AI agents.

8. As of October 2026

   Human-like agents in physical spaces

   As of October 2026, she is LiveX AI's co-founder, president, and chief AI officer. In February 2026, she describes how human-like AI holograms bridge the digital and physical worlds. At a Stanford event in May 2026, she appears as LiveX's co-founder.