Pete Warden

Career

Pete Warden describes Warden's own work as that of an engineer, CEO of Useful Sensors Inc, a founding member of the TensorFlow team, and author of TinyML. Warden is a co-author of both the 2016 TensorFlow white paper on large-scale machine learning across heterogeneous distributed systems and the 2016 TensorFlow system paper.¹²³

Research and open-source work

In 2018 Warden published the Speech Commands dataset paper as sole author, describing a dataset for training and evaluating keyword-spotting models. With Useful Sensors colleagues, Warden co-authored the 2024 Moonshine paper on speech recognition for live transcription and voice commands and the 2026 Moonshine v2 paper on a streaming encoder for latency-critical speech applications.⁸⁵⁶

The Moonshine repository says its models run entirely on-device and publishes the code and most models under the MIT License. In 2026 Warden's blog announced Moonshine Micro, aimed at low-cost embedded chips, and described porting Moonshine to JavaScript so it can run inside a web page.⁷⁹¹⁰

Teaching and writing

Warden co-wrote TinyML with Daniel Situnayake, a practical book on training models small enough for microcontrollers. Its projects include a speech recognizer, a camera that detects people and a gesture-controlled magic wand, built with TensorFlow Lite for Microcontrollers, and a series of screencasts working through the tutorial chapters is available on YouTube.⁴

Work and contributions

• TensorFlow Open-source code

  Founding member of the TensorFlow team and co-author of the TensorFlow white paper and of the 2016 paper TensorFlow: A system for large-scale machine learning.¹²³

• Moonshine Open weights

  Co-author of the 2024 Moonshine paper on speech recognition models for live transcription and voice commands and of the 2026 Moonshine v2 paper on streaming recognition for edge devices; the repository says everything runs on-device and licenses code and most models under MIT.⁵⁶⁷

• TinyML (book) Published research

  Co-author, with Daniel Situnayake, of TinyML, a book on training models small enough for microcontrollers that covers TensorFlow Lite for Microcontrollers.⁴¹

• Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition Open data

  Sole author of the 2018 paper describing the Speech Commands dataset for training and evaluating limited-vocabulary speech recognition models.⁸

Timeline

1. Mar 14, 2016Co-authored the TensorFlow white paper on large-scale machine learning on heterogeneous distributed systems.²
2. May 27, 2016Co-authored "TensorFlow, A system for large-scale machine learning," the TensorFlow system paper.³
3. Apr 9, 2018Published the Speech Commands dataset paper as sole author.⁸
4. Oct 21, 2024Co-authored the Moonshine paper on speech recognition for live transcription and voice commands.⁵
5. Feb 12, 2026Co-authored the Moonshine v2 paper on an ergodic streaming encoder for speech recognition.⁶
6. Jul 14, 2026Announced Moonshine Micro on Warden's blog.⁹
7. Aug 15, 2026Described porting Moonshine to JavaScript for in-browser speech recognition.¹⁰

Other documented roles

Roles the sources document that are not confirmed as current.

• Founding member of the TensorFlow team, Google (TensorFlow team)¹²³

Where they work

• Useful Sensors Inc 1 person in the wiki

Related people

• Kaichao YouInference and serving · ML frameworks and libraries
• Lin QiaoInference and serving · ML frameworks and libraries
• Shinji WatanabeSpeech and audio · ML frameworks and libraries
• Tim DavisML frameworks and libraries · Inference and serving
• Gabriel SynnaeveSpeech and audio · ML frameworks and libraries
• Patrick von PlatenML frameworks and libraries · Inference and serving
• Ian BuckML frameworks and libraries · Inference and serving
• Jonathan CohenSpeech and audio · ML frameworks and libraries
• Chris LattnerML frameworks and libraries · Inference and serving
• Gavin UbertiInference and serving · ML frameworks and libraries