Edge AI · Available Now · Self-Paced

Build Production Edge AI Pipeline

EdgeAI from fundamentals to advanced, in one build. Start with raw sensor data on an ESP32 or STM32 — leave having trained, quantized, and deployed a model to a Raspberry Pi or Jetson, with live predictions running on a real dashboard.

15 hoursDuration
5Modules
4Projects
Self-pacedFormat

Curriculum

Wire a sensor to the ESP32, read raw values, and build a clean preprocessing pipeline ready for model input.

Why Edge AI: cloud inference vs on-device35 min
Embedded systems crash course45 min
Reading sensor data on ESP3255 min
Project checkpoint: clean sensor pipeline45 min

Train a model on collected sensor data, then quantize it to int8 for on-device deployment.

Preparing a training dataset45 min
Training your first edge model1 hr 10 min
Understanding int8 quantization55 min
Project checkpoint: validated model1 hr 10 min

Deploy the quantized model to a Raspberry Pi and validate on-device accuracy against your training results.

TFLite Micro vs full TFLite35 min
Deploying the model to Raspberry Pi / Jetson55 min
Running live on-device inference45 min
Project checkpoint: edge inference45 min

Stream predictions over MQTT and build a real-time dashboard to visualize live results.

MQTT fundamentals for IoT telemetry40 min
Publishing predictions in real time45 min
Building a live dashboard55 min
Project checkpoint: live predictions40 min

Handle edge cases, add basic OTA update support, and review the system end to end.

Handling edge cases and failure modes30 min
OTA firmware updates30 min
Power and performance optimization30 min
Final review: end-to-end walkthrough30 min
From raw input to deployed system

What you'll build

This course follows one complete system from start to finish — not a series of disconnected exercises. By the end, you'll have a real, deployed pipeline: a sensor feeding data through a trained and quantized model, running live inference on hardware, with predictions visible on a dashboard.

Every module builds directly on the last. Each lesson ends with something working, and every project is a checkpoint toward the final system.

01Sensor data pipeline on real embedded hardware
02Trained, quantized model sized for edge deployment
03Live on-device inference, not cloud-dependent
04Working dashboard with real-time predictions
Analog Data subscription

One subscription. Every course. Real projects.

Included with an Analog Data subscription. Includes the course, linked projects, source code, and future updates.

Start Building

FAQ

The course can be followed without hardware, but to experience the hands-on build you will need the relevant components, such as an ESP32 or STM32 board, sensors, and a Raspberry Pi or Jetson. Hardware is purchased separately from different online sellers, so you can choose the exact boards and components that fit your setup.

Your Analog Data subscription unlocks this course, every other course, and the linked projects as they are published. It is the only way to access the course library.

You need basic Python, not embedded experience. The course explains every hardware concept from first principles as you go.

Learn from the builder

Your instructor

Rajath Kumar K S
Rajath Kumar K S
Founder, Analog Data · Bengaluru
Delivers hands-on embedded and edge AI training, including sessions for engineering teams at ISRO and Broadridge. This course is built from the same curriculum used in that live training.
Build with Analog Data

Stop watching. Start shipping.

Real hardware. Real projects. A production Edge AI pipeline that makes your profile stand out.

Real projects not toy exercisesProduction context from start to finish
Start Building