Microsoft
Australian National University
Stanford University
Oriol VinyalsCo-founder · Discovery LoopResearcher↗
Researcher · Founder
Quoc V. Le
Discovery LoopCo-founder
🇻🇳 VietnameseBorn 1982
Le helped found Google Brain and co-created sequence-to-sequence learning for translation, neural architecture search for designing networks, and EfficientNet for recognising images with less computing power. A Google Fellow who helped develop Gemini, he left Google DeepMind in 2026 to co-found Discovery Loop, which aims to automate scientific research.[1]
Career & education
Current roleDiscovery Loop2026-presentCo-founder
Discovery Loop2026-presentPreviously
Google DeepMindUntil 2026Google Fellow
Google DeepMindUntil 2026Work and education history is primarily focused on AI-relevant roles and may not be comprehensive.
Citation Trend
Total citations to scholarly work, including research papers, technical reports, books and theses, as recorded by Google Scholar. Each point is a dated snapshot.
citations484,137As of Oct 1, '26
citations+9,247Since last snapshot, Sep 1, '26
Updates 2
MoveQuoc Le moves from Google DeepMind to co-found Discovery Loop
The Google Brain founding researcher and Google Fellow is co-founding Discovery Loop with three longtime collaborators.
August 5, 2026LaunchJeff Dean, Oriol Vinyals, Quoc Le, and Sanjay Ghemawat launch Discovery LoopFour longtime Google collaborators founded a public benefit corporation to automate machine learning, science, and engineering.
August 5, 2026Selected contributions
Sequence to Sequence Learning with Neural Networks
With Ilya Sutskever and Oriol VinyalsShowed how two neural networks could turn an input sequence into an output sequence of a different length, achieving strong results translating English into French.
Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
Introduced a way to build much larger neural networks by routing each input to a few specialist subnetworks, increasing capacity without a proportional increase in computation.
XLNet: Generalized Autoregressive Pretraining for Language Understanding
Introduced XLNet, a language model trained to predict words in varied orders so it can learn context from both sides of a sentence. It improved on BERT across tasks including question answering, sentiment analysis, and document ranking without relying on artificial masked words during training.
References
- Discovery Loop launch interviewwired.com
August 2026 Google departure, founding team and aim to automate scientific research.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed July 19, 2026 · Checked for accuracy and new information.
