Quoc V. Le
Google Fellow at Google DeepMind
A Google Brain founding member whose work on sequence-to-sequence models, AutoML and chain-of-thought prompting shaped modern language and vision systems.
Biography
Quoc Viet Le is a Vietnamese-American computer scientist and Google Fellow at Google DeepMind. Born in Hương Thủy, Vietnam, he moved to Australia in 2004 for a bachelor's degree at the Australian National University, then to Stanford University in 2007, where Andrew Ng supervised his PhD.
In 2011 he became a founding member of Google Brain with Ng, Jeff Dean and Greg Corrado, and led its first major breakthrough, a network trained on 16,000 CPU cores that learned to recognize cats from YouTube videos. In 2014 he co-introduced sequence-to-sequence learning with Ilya Sutskever and Oriol Vinyals, and developed the doc2vec document embedding with Tomáš Mikolov. He started Google's AutoML project in 2017, which led to EfficientNet, and he contributed to Meena (later LaMDA), chain-of-thought prompting and AlphaGeometry. Wikipedia's entry on Oriol Vinyals reports that in August 2026 Vinyals announced a new company, Discovery Loop, with Le, Jeff Dean and Sanjay Ghemawat.
Notable contributions
Co-introduced sequence-to-sequence learning
With Ilya Sutskever and Oriol Vinyals, proposed a neural approach that maps an input sequence to an output sequence, demonstrated on machine translation.
Led Google's AutoML and neural architecture search
Started the AutoML project at Google Brain, pioneering neural architecture search, which produced the EfficientNet family of smaller, faster image recognition models.
Co-authored chain-of-thought prompting
With coauthors, published a method that improves the reasoning of large language models by prompting them to write out intermediate steps.
Helped found Google Brain
Became a founding member of the Google Brain project in 2011 and led its first major result, a network that learned to recognize cats from unlabelled YouTube videos.
