# Barret Zoph

> Source: https://aiwiki.ai/wiki/barret_zoph
> Updated: 2026-10-27
> Fact-checked: 2026-11-05
> Categories: AI Research, People
> License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)"
> Cite as: AI Wiki. "Barret Zoph." aiwiki.ai, 27 Oct 2026. https://aiwiki.ai/wiki/barret_zoph
> From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution.

**Barret Zoph** is an American [artificial intelligence](https://aiwiki.ai/wiki/artificial_intelligence) researcher and engineer best known for pioneering [neural architecture search](https://aiwiki.ai/wiki/neural_architecture_search) (NAS) at [Google Brain](https://aiwiki.ai/wiki/google_brain) and for helping build [ChatGPT](https://aiwiki.ai/wiki/chatgpt) as a vice president of research at [OpenAI](https://aiwiki.ai/wiki/openai) [1][2]. After leaving OpenAI in 2024 he co-founded [Thinking Machines Lab](https://aiwiki.ai/wiki/thinking_machines_lab) with former OpenAI chief technology officer [Mira Murati](https://aiwiki.ai/wiki/mira_murati), serving as the startup's co-founder and chief technology officer, before returning to OpenAI in January 2026 to lead the company's enterprise business [3][4][5]. In August 2026 he said he would join [Google DeepMind](https://aiwiki.ai/wiki/google_deepmind) as vice president of research [27][28].

## Overview

Zoph's research spans automated [machine learning](https://aiwiki.ai/wiki/machine_learning), large-scale language modeling, and the [post-training](https://aiwiki.ai/wiki/post-training) methods used to align and adapt [foundation models](https://aiwiki.ai/wiki/foundation_models). His early work with [Quoc V. Le](https://aiwiki.ai/wiki/quoc_le) framed the design of [neural network](https://aiwiki.ai/wiki/neural_network) architectures as a learning problem in its own right, producing the NAS and NASNet methods that influenced a generation of [AutoML](https://aiwiki.ai/wiki/automl) research [6][7]. He also became one of the principal researchers behind sparse [mixture of experts](https://aiwiki.ai/wiki/mixture_of_experts) [language model](https://aiwiki.ai/wiki/language_model)s, co-authoring the [Switch Transformer](https://aiwiki.ai/wiki/switch_transformer) and several follow-up papers on training large sparse models stably [14][22]. At OpenAI he co-led the team responsible for turning raw language models into the conversational system released as ChatGPT [1][8]. His publications are widely cited, according to his [Google Scholar](https://scholar.google.com/citations?user=NL_7iTwAAAAJ&hl=en) profile [2].

## Education

Zoph earned a Bachelor of Science in computer science from the University of Southern California (USC) in 2016 [9]. As an undergraduate he worked with USC faculty members Kevin Knight and David Kempe on computer science research [9]. He has said that getting involved in research as an undergraduate, and finding the right faculty mentor, made a large difference to his career, and that learning how to build the infrastructure for and train [deep learning](https://aiwiki.ai/wiki/deep_learning) systems became a skill he relied on throughout his later work [9]. Before joining [Google](https://aiwiki.ai/wiki/google) he carried out research at USC's Information Sciences Institute (ISI), collaborating with Kevin Knight and Daniel Marcu on statistical [machine translation](https://aiwiki.ai/wiki/machine_translation), a field for which ISI was well known [1][9].

That ISI work produced one of his first widely cited papers, "Transfer Learning for Low-Resource Neural Machine Translation," presented at EMNLP 2016 with Deniz Yuret, Jonathan May, and Kevin Knight. The method first trained a high-resource language pair to create a parent model, then transferred some of its learned parameters to initialize and constrain training on a low-resource pair, improving baseline neural translation systems by an average of 5.6 BLEU across four low-resource language pairs [23]. After graduating in 2016 he joined Google Brain, entering through the Google Brain Residency program [30].

## Google Brain and neural architecture search

At Google Brain, Zoph worked as a research scientist, later a staff research scientist, on automated machine learning and, subsequently, on large sparse language models [1]. His best-known contribution from this period is neural architecture search. In the 2017 paper "Neural Architecture Search with Reinforcement Learning," written with Quoc V. Le, he used a [recurrent neural network](https://aiwiki.ai/wiki/recurrent_neural_network) controller, trained with [reinforcement learning](https://aiwiki.ai/wiki/reinforcement_learning), to generate the descriptions of candidate networks and to maximize their validation accuracy [7]. On the [CIFAR-10](https://aiwiki.ai/wiki/cifar_10) image dataset the method discovered architectures competitive with the best hand-designed models of the time, and on the Penn Treebank language-modeling benchmark it produced a recurrent cell that outperformed the standard LSTM [7].

Zoph and collaborators extended the idea in "Learning Transferable Architectures for Scalable Image Recognition" (2018), which introduced the NASNet design. Rather than search over a whole network, the method searched for a reusable convolutional cell on a small dataset and then stacked copies of that cell to build larger models, allowing an architecture found on CIFAR-10 to transfer to the much larger [ImageNet](https://aiwiki.ai/wiki/imagenet) benchmark [11]. He also contributed to data-augmentation research, co-authoring AutoAugment and RandAugment, which learn augmentation policies from data [12][13], and to work on large mixture of experts language models. His architecture-search ideas fed into Google's AutoML products and influenced later efficient-network research such as [EfficientNet](https://aiwiki.ai/wiki/efficientnet) [6][11].

In his later years at Google, Zoph focused on training large sparse models. With William Fedus and [Noam Shazeer](https://aiwiki.ai/wiki/noam_shazeer) he co-authored the Switch Transformer, which simplified mixture-of-experts routing so that each token was sent to a single expert, allowing models to scale to a trillion parameters while keeping the computational cost per token roughly constant and reaching up to seven times faster pre-training than dense baselines of comparable cost [14]. He followed this with "ST-MoE: Designing Stable and Transferable Sparse Expert Models" (2022), which addressed the training instability and fine-tuning difficulties of sparse models and scaled a sparse model to 269 billion parameters at a compute cost comparable to a 32-billion-parameter dense Transformer, the first time a sparse model reached state-of-the-art transfer-learning results across a broad set of language tasks [22].

### Selected works

| Year | Title | Co-authors | Contribution |
| --- | --- | --- | --- |
| 2016 | [Transfer Learning](https://aiwiki.ai/wiki/transfer_learning) for Low-Resource [Neural Machine Translation](https://aiwiki.ai/wiki/neural_machine_translation) | D. Yuret, J. May, K. Knight | Parent-to-child parameter transfer for low-resource translation; +5.6 BLEU on average [23] |
| 2017 | [Neural Architecture Search](https://aiwiki.ai/wiki/neural_architecture_search) with [Reinforcement Learning](https://aiwiki.ai/wiki/reinforcement_learning) | [Quoc V. Le](https://aiwiki.ai/wiki/quoc_le) | RNN controller trained with RL to design networks; founded NAS [7] |
| 2018 | Learning Transferable Architectures for Scalable [Image Recognition](https://aiwiki.ai/wiki/image_recognition) | V. Vasudevan, J. Shlens, Q. V. Le | Introduced the NASNet transferable-cell design [11] |
| 2019 | AutoAugment: Learning Augmentation Strategies from Data | E. D. Cubuk, D. Mané, V. Vasudevan, Q. V. Le | Learned data-augmentation policies [12] |
| 2020 | RandAugment: Practical Automated [Data Augmentation](https://aiwiki.ai/wiki/data_augmentation) | E. D. Cubuk, J. Shlens, Q. V. Le | Simplified, reduced-search augmentation [13] |
| 2022 | [Switch Transformer](https://aiwiki.ai/wiki/switch_transformer)s | W. Fedus, N. Shazeer | Trillion-parameter sparse mixture-of-experts model [14] |
| 2022 | ST-MoE: Designing Stable and Transferable Sparse Expert Models | I. Bello, S. Kumar, N. Du, Y. Huang, J. Dean, N. Shazeer, W. Fedus | Stable 269B-parameter sparse model with state-of-the-art transfer [22] |
| 2023 | GPT-4 Technical Report | [OpenAI](https://aiwiki.ai/wiki/openai) | Contributor to OpenAI's flagship model [2][8] |

## OpenAI and ChatGPT

Zoph joined OpenAI in September 2022, shortly before the public launch of ChatGPT [9][15]. He helped build the company's post-training team from scratch alongside [John Schulman](https://aiwiki.ai/wiki/john_schulman) and others, and rose to vice president of research for post-training [1][15]. The post-training group is responsible for the work that turns a pretrained [large language model](https://aiwiki.ai/wiki/large_language_model) into a usable assistant, including instruction following, [reinforcement learning from human feedback](https://aiwiki.ai/wiki/rlhf), tool use, evaluations, and safety filtering [1][8]. Zoph's teams trained the models shipped into ChatGPT and the [OpenAI API](https://aiwiki.ai/wiki/openai_api), and his team's work extended across alignment, tool use, evaluations, search, and multimodality [1]; he was described as a post-training co-lead for [GPT-4](https://aiwiki.ai/wiki/gpt-4) [15]. He is credited as a contributor on the GPT-4 Technical Report [2][8].

On January 28, 2025, after both had left the company, Zoph and Schulman gave a Stanford HAI seminar titled "ChatGPT and the Art of Post Training" on the early development of ChatGPT's post-training [24].

## Departure and Thinking Machines Lab

On September 25, 2024, Zoph left OpenAI, announcing his departure the same week that Mira Murati said she would step down as chief technology officer, and chief research officer [Bob McGrew](https://aiwiki.ai/wiki/bob_mcgrew) said he would leave [15][16]. In a note to OpenAI staff that he also posted publicly, Zoph called leaving "a very difficult decision" and "a personal decision," saying it felt like "a natural point" to explore new opportunities outside OpenAI [15][16]. Chief executive [Sam Altman](https://aiwiki.ai/wiki/sam_altman) said the three had made their decisions "independently of each other and amicably," with the exits announced together for a smooth handover [16].

In February 2025, Zoph re-teamed with Murati as a co-founder of Thinking Machines Lab, an AI startup incorporated as a public benefit corporation with the stated goal of making AI systems more widely understood, customizable, and capable [17][18]. The founding group also included John Schulman as chief scientist, [Lilian Weng](https://aiwiki.ai/wiki/lilian_weng), [Andrew Tulloch](https://aiwiki.ai/wiki/andrew_tulloch), and Luke Metz, most of them former OpenAI researchers [17][18]. Zoph served as the company's chief technology officer [3][18]. In July 2025 the startup raised about $2 billion at a roughly $12 billion valuation in a round led by [Andreessen Horowitz](https://aiwiki.ai/wiki/andreessen_horowitz), with participation from investors including [Nvidia](https://aiwiki.ai/wiki/nvidia), AMD, Cisco, and Jane Street, and in October 2025 it shipped its first product, a model fine-tuning service called [Tinker](https://aiwiki.ai/wiki/tinker) [18][19].

## Return to OpenAI

On January 14, 2026, OpenAI announced that Zoph would return to the company along with Luke Metz and Sam Schoenholz, all three departing Thinking Machines Lab [4][20]. [Fidji Simo](https://aiwiki.ai/wiki/fidji_simo), OpenAI's chief executive of applications, said she was "excited to welcome Barret Zoph, Luke Metz, and Sam Schoenholz back to OpenAI," adding that the move had "been in the works for several weeks" [4][20]. Announcing the split roughly an hour before OpenAI confirmed the returns, Murati said the company had "parted ways with Barret Zoph" and named [Soumith Chintala](https://aiwiki.ai/wiki/soumith_chintala), the creator of [PyTorch](https://aiwiki.ai/wiki/pytorch), as the new chief technology officer of Thinking Machines Lab [4][25]. Wired reported that the separation was not amicable. Citing Simo's memo, it said Zoph told Murati on the Monday that he was considering leaving and was fired on the Wednesday. Technology reporter Kylie Robison wrote that he was fired for "unethical conduct," and a source close to Thinking Machines alleged that he had shared confidential company information with competitors. Wired said it was unable to verify this with Zoph, and Simo wrote that OpenAI did not share Murati's concerns [4][25][26][29]. Coverage of the move noted that Zoph's return came as OpenAI reorganized around enterprise growth, with Zoph reporting to Simo and taking on leadership of the company's enterprise and commercial business rather than a purely research role [5][21][25].

## Google DeepMind

On August 26, 2026, Zoph said he would join Google DeepMind as vice president of research, Reuters reported, noting that his return to OpenAI had been brief, according to his LinkedIn profile [27]. The move took him back to the organization where he had spent six years as a research scientist at Google Brain, which merged with DeepMind in 2023 to form Google DeepMind [27][28]. In his announcement Zoph said he would work on reinforcement learning and post-training [30]. The Next Web noted that no remit or team had been published [28].

## Recognition

His 2017 neural architecture search paper alone has accumulated thousands of citations [2][7]. His architecture-search and sparse-model work is frequently cited as foundational to AutoML and to the design of large mixture-of-experts language models [6][22]. USC has highlighted him as an alumnus whose work helped pave the way for ChatGPT [9].

## References

1. Barret Zoph, personal website. https://barretzoph.github.io/
2. Barret Zoph, Google Scholar profile. https://scholar.google.com/citations?user=NL_7iTwAAAAJ&hl=en
3. "Inside Thinking Machines Lab, Mira Murati's New AI Startup," Built In. https://builtin.com/articles/what-is-thinking-machines-lab
4. "Mira Murati's startup, Thinking Machines Lab, is losing two of its co-founders to OpenAI," TechCrunch, January 14, 2026. https://techcrunch.com/2026/01/14/mira-muratis-startup-thinking-machines-lab-is-losing-two-of-its-co-founders-to-openai/
5. "OpenAI is coming for those sweet enterprise dollars in 2026," TechCrunch, January 22, 2026. https://techcrunch.com/2026/01/22/openai-is-coming-for-those-sweet-enterprise-dollars-in-2026/
6. "Neural architecture search," Wikipedia. https://en.wikipedia.org/wiki/Neural_architecture_search
7. Barret Zoph and Quoc V. Le, "Neural Architecture Search with Reinforcement Learning," arXiv:1611.01578 (ICLR 2017). https://arxiv.org/abs/1611.01578
8. OpenAI, "GPT-4 Technical Report," arXiv:2303.08774. https://arxiv.org/abs/2303.08774
9. "Attention Is All You Need: USC Alumni Paved Path for ChatGPT," USC Information Sciences Institute, 2023. https://www.isi.edu/news/54564/attention-is-all-you-need-usc-alumni-paved-path-for-chatgpt/
10. "Neural Architecture Search with Reinforcement Learning," Google Research. https://research.google/pubs/neural-architecture-search-with-reinforcement-learning/
11. Barret Zoph, Vijay Vasudevan, Jonathon Shlens, and Quoc V. Le, "Learning Transferable Architectures for Scalable Image Recognition," arXiv:1707.07012 (CVPR 2018). https://arxiv.org/abs/1707.07012
12. Ekin D. Cubuk, Barret Zoph, Dandelion Mané, Vijay Vasudevan, and Quoc V. Le, "AutoAugment: Learning Augmentation Strategies from Data," CVPR 2019. https://openaccess.thecvf.com/content_CVPR_2019/papers/Cubuk_AutoAugment_Learning_Augmentation_Strategies_From_Data_CVPR_2019_paper.pdf
13. Ekin D. Cubuk, Barret Zoph, Jonathon Shlens, and Quoc V. Le, "RandAugment: Practical Automated Data Augmentation with a Reduced Search Space," arXiv:1909.13719. https://arxiv.org/abs/1909.13719
14. William Fedus, Barret Zoph, and Noam Shazeer, "Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity," Journal of Machine Learning Research, 2022. https://arxiv.org/abs/2101.03961
15. "Barret Zoph, GPT-4's post-training (co-)lead is also leaving," Hacker News discussion. https://news.ycombinator.com/item?id=41653251
16. "OpenAI's chief research officer has left following CTO Mira Murati's exit," TechCrunch, September 25, 2024. https://techcrunch.com/2024/09/25/openais-chief-research-officer-has-left/
17. "Thinking Machines Lab is ex-OpenAI CTO Mira Murati's new startup," TechCrunch, February 18, 2025. https://techcrunch.com/2025/02/18/thinking-machines-lab-is-ex-openai-cto-mira-muratis-new-startup/
18. "Thinking Machines Lab," Wikipedia. https://en.wikipedia.org/wiki/Thinking_Machines_Lab
19. "Former OpenAI CTO Mira Murati raises $2 billion for new AI startup Thinking Machines Lab," CNBC, July 15, 2025. https://www.cnbc.com/2025/07/15/openai-mira-murati-thinking-machines-lab.html
20. "Thinking Machines Lab CTO Barret Zoph returns to OpenAI in surprise move," Neowin, January 2026. https://www.neowin.net/news/thinking-machines-lab-cto-barret-zoph-returns-to-openai-in-surprise-move/
21. "Barret Zoph To Lead OpenAI's Aggressive Commercial Pivot," Dataconomy, January 23, 2026. https://dataconomy.com/2026/01/23/barret-zoph-to-lead-openais-aggressive-commercial-pivot/
22. Barret Zoph, Irwan Bello, Sameer Kumar, Nan Du, Yanping Huang, Jeff Dean, Noam Shazeer, and William Fedus, "ST-MoE: Designing Stable and Transferable Sparse Expert Models," arXiv:2202.08906 (2022). https://arxiv.org/abs/2202.08906
23. Barret Zoph, Deniz Yuret, Jonathan May, and Kevin Knight, "Transfer Learning for Low-Resource Neural Machine Translation," Proceedings of EMNLP 2016. https://aclanthology.org/D16-1163/
24. John Schulman and Barret Zoph, "ChatGPT and the Art of Post Training," Stanford HAI seminar, January 28, 2025. https://hai.stanford.edu/events/john-schulman-and-barret-zoph-chatgpt-and-the-art-of-post-training
25. "Thinking Machines Ousts CTO Zoph; Three Return to OpenAI," The Implicator, January 2026. https://www.implicator.ai/thinking-machines-ousts-cto-barret-zoph-hours-later-openai-welcomed-him-home/
26. "A source close to Thinking Machines Lab alleges ex-CTO Barret Zoph had shared confidential company information with competitors," Wired (via Techmeme), January 15, 2026. https://www.techmeme.com/260115/p2
27. "Thinking Machines Lab co-founder Barret Zoph joins Google," Reuters (via The Star), August 27, 2026. https://www.thestar.com.my/tech/tech-news/2026/08/27/thinking-machines-lab-co-founder-barret-zoph-joins-google
28. "Barret Zoph is going back to Google, six years and two startups later," The Next Web, August 27, 2026. https://thenextweb.com/news/barret-zoph-google-deepmind-thinking-machines
29. "Two Thinking Machines Lab Cofounders Are Leaving to Rejoin OpenAI," Wired, January 14, 2026. https://www.wired.com/story/thinking-machines-lab-cofounders-leave-for-openai/
30. "Thinking Machines Co-founder Barrett Zoph Leaves OpenAI To Join Google," OfficeChai, August 2026. https://officechai.com/ai/thinking-machines-co-founder-barrett-zoph-leaves-openai-to-join-google/