Deploy EdgeAI Applications with ESP32-S3: A Hands-On TinyML Workshop

Run speech and vision inference using TensorFlow Lite Micro, Espressif AI frameworks, and Edge Impulse in an ESP-IDF simulation environment.

Edge AI adoption is accelerating across IoT and embedded systems, and TinyML enables machine learning inference to run directly on microcontrollers. Instead of sending data to the cloud, models can now execute locally on devices such as the ESP32-S3, which is Espressif’s high-performance MCU designed for AIoT applications.

While many engineers are familiar with machine learning concepts, integrating real-world inference into embedded firmware presents practical challenges.

This workshop bridges that gap.

Led by Vedat Ozan Oner, Computer Engineer, IoT product architect, visiting lecturer, and author of blockbuster series Developing IoT Projects with ESP32 (Packt, 1st and 2nd editions), this workshop draws on over 20 years of embedded systems expertise to transform complex TinyML and ESP32 integration challenges into clear, practical, firmware-level deployment strategies grounded in real-world product development.

In this intensive, hands-on session, you will deploy and run real TinyML inference workflows targeting the ESP32-S3 architecture using:

• TensorFlow Lite for Microcontrollers (TFLM)

• Espressif Artificial Intelligence of Things (AIoT) frameworks

• Edge Impulse Platform

All examples are built with ESP-IDF and executed in a fully simulated environment using Wokwi.

No ESP32 board required. No hardware purchase needed.

You will leave with a working, reproducible development environment and a clear architectural understanding of how TinyML solutions are structured inside embedded firmware.

This is not a machine learning theory session. It is a firmware-level integration workshop designed for engineers who want practical deployment experience.

By the End You’ll Walk Away With:

• The ability to run and analyze TinyML inference workloads on ESP32-S3

• Practical, hands-on experience running speech and vision inference models using TFLM

• Confidence working with Espressif AIoT frameworks and libraries for embedded AI applications

• Experience deploying and executing models using the Edge Impulse C++ SDK

• A clear understanding of the abstraction differences between TensorFlow Lite Micro, Espressif AIoT frameworks, and Edge Impulse

• A fully working ESP-IDF + Wokwi TinyML simulation environment you can reuse for your own projects

• Reusable scaffold code and configurations for your own projects

Who Should Attend:

• Embedded firmware engineers working with ESP-IDF

• IoT developers exploring on-device ML integration

• Engineers transitioning from cloud ML to edge ML

• Makers and developers serious about ESP32-based ML applications

• This is also useful for engineering students building Embedded ML portfolios if they have basic knowledge of embedded systems, C/C++ or Python, and microcontrollers.

This workshop assumes programming experience and comfort working with developer tools.

It is not a fundamentals-only or demo-driven session.

It is designed for developers who want real TinyML integration experience.

What You’ll Learn:

• How to run TinyML inference on ESP32-S3 using ESP-IDF in a Wokwi simulation environment

• The core components of a TinyML application built with TFLM

• How to integrate and test speech and vision models using TFLM with offline input data

• How to work with Espressif AIoT frameworks (ESP-SR, ESP-WHO, ESP-SKAINET, ESP-DL, ESP-NN) and understand their use-cases

• How to develop wake-word and face detection applications using Espressif AIoT frameworks

• How the Edge Impulse Platform supports the TinyML development workflow

• How to deploy models using the Edge Impulse C++ SDK

• The key architectural differences between three TinyML deployment approaches

What You’ll Need & Prerequisites:

• A laptop (Windows, macOS, or Linux)
• Docker installed and running
• Visual Studio Code
• Git installed
• A Wokwi account (Community license)
• Stable internet connection
• Basic knowledge of C or C++ is required.
• Familiarity with microcontroller development concepts is helpful
• No prior machine learning experience is required
• No physical ESP32 hardware is required. All exercises are completed in simulation

A scaffold repository with a preconfigured ESP-IDF development container will be provided.

Why Now?

Edge AI adoption is accelerating across IoT and embedded products.

The ESP32-S3 introduces AI acceleration features, and more teams are expected to integrate on-device inference into firmware.

However, many engineers lack practical deployment experience across frameworks.

This workshop provides a structured, reproducible path to understanding how TinyML actually runs inside embedded systems before moving to physical hardware or production design.

🎟️ Reserve your spot now

*By signing up for this event, you agree to recieving emails from Packt Publishing.