Ashish Vaswani

Research

Ashish Vaswani is the first-listed author of "Attention Is All You Need" (2017), written with a Google Brain affiliation; the paper's contribution note says Vaswani, with Illia Polosukhin, designed and implemented the first Transformer models. In 2018 Vaswani was first author of the Tensor2Tensor paper, describing the Google Brain library that holds the reference Transformer implementation.⁵⁶⁷

Vaswani also co-authored several 2018 papers extending self-attention: relative position representations with Peter Shaw and Jakob Uszkoreit, the Image Transformer for image generation, and the Music Transformer for generating music with long-term structure.⁸⁹¹⁰

Essential AI

As CEO of Essential AI, Vaswani is listed among the authors of the company's 2025 research papers, including "Rethinking Reflection in Pre-Training" (April 2025), "Practical Efficiency of Muon for Pretraining" (May 2025) and the Essential-Web v1.0 dataset paper (June 2025).¹¹¹¹²⁴

In December 2025 Vaswani wrote the announcement of Rnj-1, an 8B base and instruction-tuned model pair that roughly follows the Gemma 3 architecture and uses YaRN to extend context to 32k tokens. The models are published on Hugging Face under the Apache 2.0 license.²³

Work and contributions

• Rnj-1 Open weights

  Author of Essential AI's December 2025 announcement of Rnj-1, an 8B base and instruction-tuned model pair that roughly follows the Gemma 3 architecture; the Hugging Face model card lists the Apache 2.0 license.²³

• Essential-Web v1.0 Open data

  Co-author of the 2025 Essential-Web v1.0 paper, which presents a 24-trillion-token dataset in which every document is annotated with a taxonomy; the dataset card lists the ODC-By license.⁴¹³

• Transformer (Attention Is All You Need) Published research

  First-listed author of the 2017 paper that introduced the Transformer; the paper lists the author order as random and says Vaswani, with Illia Polosukhin, designed and implemented the first Transformer models.⁵

• Tensor2Tensor Open-source code

  First author of the 2018 Tensor2Tensor paper; the Apache-2.0 library, developed by the Google Brain team, includes the reference implementation of the Transformer.⁶⁷

• Rnj-1 Instruct Open weights

  First listed author in the citation block of the Rnj-1 Instruct model card; Essential AI's Rnj-1 announcement credits Vaswani in the leadership group for the research and engineering roadmap.¹⁴²

• Rnj-1.5 Instruct Open weights

  Listed among the authors in the citation block of the Rnj-1.5 Instruct model card.¹⁵

Timeline

1. Jun 12, 2017Posted "Attention Is All You Need" as first-listed author, introducing the Transformer.⁵
2. Feb 15, 2018Co-authored the Image Transformer paper.⁹
3. Mar 6, 2018Co-authored "Self-Attention with Relative Position Representations" with Peter Shaw and Jakob Uszkoreit.⁸
4. Mar 16, 2018Posted the Tensor2Tensor for Neural Machine Translation paper as first author.⁶
5. Sep 12, 2018Co-authored the Music Transformer paper.¹⁰
6. Apr 5, 2025Co-authored "Rethinking Reflection in Pre-Training" with Essential AI colleagues.¹¹
7. May 4, 2025Co-authored "Practical Efficiency of Muon for Pretraining" with Essential AI colleagues.¹²
8. Jun 17, 2025Co-authored the Essential-Web v1.0 paper on a 24-trillion-token organized web dataset.⁴
9. Dec 5, 2025Announced Essential AI's Rnj-1 base and instruction-tuned models.²

Other documented roles

Roles the sources document that are not confirmed as current.

• Listed with a Google Brain affiliation on 2017–2018 papers, Google Brain⁵⁶

Where they work

• Essential AI 1 person in the wiki

Related people

• Albert JiangLanguage models · Datasets and data
• Guilherme PenedoDatasets and data · Language models
• John X. MorrisLanguage models · Datasets and data
• Sridhar RamaswamyLanguage models · Datasets and data
• Hynek KydlíčekDatasets and data · Language models
• Yuanzhi LiLanguage models · Datasets and data
• Rangan MajumderLanguage models · Datasets and data
• Aran KomatsuzakiLanguage models · Datasets and data
• Henrique Schechter VeraLanguage models · Datasets and data
• Sheshansh AgrawalLanguage models · Datasets and data