No. 353 of 500 No. 51 in Hands-On AI & Programming

TinyML

Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers

By Pete Warden and Daniel Situnayake

Print length
504 pages
Publisher
O'Reilly Media
Reading level
Intermediate
ISBN-13
978-1492052043
Published
January 21, 2020
Written for
Embedded engineers, makers and software developers who want to deploy machine learning on microcontrollers.

Synopsis

Pete Warden and Daniel Situnayake, who helped build TensorFlow Lite for Microcontrollers at Google, wrote this hands-on introduction to running neural networks on chips with only kilobytes of memory and very small power budgets. It is aimed at software and hardware developers and does not assume machine learning experience. The opening chapters build a minimal "hello world" model that approximates a sine wave: readers train it in TensorFlow, convert and shrink it for TensorFlow Lite, and deploy it to boards such as the Arduino Nano 33 BLE Sense and SparkFun Edge, with each step of the toolchain explained.

Later projects grow more ambitious: a wake-word detector that listens for "yes" and "no", a camera-based person detector, and a "magic wand" that recognizes gestures from accelerometer data. The final chapters cover the work of moving from prototype to product, including optimizing latency and power consumption, shrinking model size, debugging embedded applications, porting models from TensorFlow, and privacy and security safeguards. Readers come away understanding the full TinyML workflow and why on-device inference matters for battery life, privacy and cost. Because TensorFlow Lite Micro and its tooling have changed since 2019, some code and setup instructions need adapting, and the projects are teaching examples rather than production-grade models.

Scorecard

Editorial score

How it scored

Editorial score: 74.5 / 1005 weighted criteria

  • Influence30%

    22.5/30
  • Authority25%

    22.5/25
  • Practical value20%

    15.0/20
  • Currency15%

    7.5/15
  • Readability10%

    7.0/10
  • Editorial score

    74.5/100

How the rank was computed

Rank No. 353 of 500 comes from a composite of 51.5: observable signals 17.9 of 55 points plus editorial review 33.6 of 45 points.

All 11 inputs
InputPoints
Observable signals17.9/55
Reach (Amazon ratings volume)9.0/16.5
Consensus across best-of lists2.8/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted5.4/8.3
Amazon sales rank0.7/5.5
Recency0.0/2.8
Editorial review33.6/45
Author authority10.1/11.3
Influence10.1/13.5
Practical value6.8/9.0
Readability3.2/4.5
Currency3.4/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

The founding practical text for running machine learning on microcontrollers, valuable for its projects though its toolchain details have aged.

Strengths

  • Complete end-to-end projects on real hardware
  • Written by engineers who built TensorFlow Lite Micro
  • Covers optimization, debugging and deployment constraints

Keep in mind

Toolchain and library changes since 2019 mean some examples need troubleshooting, and the neural network theory is too shallow for readers wanting depth.

Key takeaways

  1. Useful neural networks can run on microcontrollers with only kilobytes of memory, enabling always-on sensing at very low power and cost.
  2. Quantization and model conversion are central TinyML skills, shrinking a TensorFlow model so it fits and runs efficiently on tiny hardware.
  3. Audio, image and accelerometer data each require different preprocessing pipelines, as the wake-word, person-detection and gesture projects demonstrate step by step.
  4. Running inference on the device itself keeps raw sensor data local, which improves privacy and battery life compared with sending data to the cloud.

Notable facts

  • An Edge AI and Vision Alliance profile describes TinyML as the standard embedded machine learning textbook and credits Pete Warden with coining the term TinyML. Source (opens in a new tab)

About the author

Pete Warden and Daniel Situnayake

Pete Warden was a founding member of Google's TensorFlow team and led development of TensorFlow Lite, including TensorFlow Lite for Microcontrollers; he is credited with coining the term TinyML and in 2022 co-founded Useful Sensors, where he is CEO. Daniel Situnayake led developer advocacy for TensorFlow Lite at Google, later became head of machine learning at Edge Impulse, and co-wrote AI at the Edge (O'Reilly, 2023).

Questions

Is TinyML still worth reading in 2026?
Yes, for its concepts and workflow, with caveats about code. The book remains one of the clearest explanations of how to train, shrink and deploy models on microcontrollers, and its projects teach durable skills in audio, vision and sensor processing. TensorFlow Lite Micro has since evolved, so expect to consult current documentation and repository examples when build steps or APIs no longer match the 2019 text.
What projects does TinyML include?
Readers build a sine-wave "hello world" model, a speech wake-word detector that recognizes the words "yes" and "no", a person detector that uses a small camera module, and a gesture-recognizing "magic wand" driven by accelerometer data. Each project runs on low-power boards such as the Arduino Nano 33 BLE Sense or SparkFun Edge, and later chapters cover optimizing, debugging and porting models.
Who should read TinyML?
It suits embedded engineers who want to add machine learning to products, software developers curious about hardware, and makers building sensor projects. Some comfort with C++, Python and the command line helps, and readers who have never programmed a microcontroller may need extra resources. Those wanting deep neural network theory or production deployment guidance should pair it with a dedicated deep learning text.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.