I taught IoT (Internet of Things) from 2005 to 2020, and during the last 3 years had integrated AI into the curriculum with the emergence of open-source AI models capable of deployment on resource-constrained microcontrollers like the ESP8266. I am now exploring the design of a comprehensive 10-week course that bridges edge computing, AI agents, and practical hardware implementation—and I’m seeking collaborators to help refine the curriculum and create the supporting materials and deployment packages. Working with AI assistance, I’ve drafted the following course outline as a starting framework:
This 10-week course outline is designed to provide a comprehensive and practical understanding of building AI-driven edge computing systems using a variety of hardware and open-source software tools.
Edge AI Agent Development Course Outline (10 Weeks)
Phase 1: Foundations and TinyML (Weeks 1-3)
| Week | Topics | Key Concepts & Labs | Hardware Focus |
|---|---|---|---|
| 1 | Introduction to Edge AI & IoT | Definition of Edge vs. Cloud, AI Agent Concepts, IoT device types, Use Cases (Industrial, Smart Home, etc.). | Overview of all devices (Pi, Jetson, ESP8266), Initial setup (OS flashing, basic connectivity). |
| 2 | Microcontroller Edge (TinyML) Fundamentals | Introduction to TinyML, ESP8266 architecture, Data acquisition from basic sensors (temp/humidity), Introduction to MicroPython/Arduino IDE. | ESP8266/ESP32 & simple sensors (DHT11/22). Lab: Sensor data reading and printing to serial. |
| 3 | TinyML Model Training and Deployment | Data collection best practices for TinyML, Training a simple ML model (e.g., keyword spotting, simple classification) using a platform like Edge Impulse. | ESP8266 (deployment), Basic data collection script. Lab: Deploying a trained TinyML model to the MCU. |
| Phase 2: Communication, Data, and Single Board Computers (Weeks 4-6) | |||
| Week | Topics | Key Concepts & Labs | Hardware Focus |
| — | — | — | — |
| 4 | Messaging Protocol: MQTT | Introduction to Pub/Sub model, MQTT broker setup (e.g., Mosquitto), Quality of Service (QoS) levels, Basic MQTT client implementation on MCU and SBC. | ESP8266, Raspberry Pi. Lab: MCU sends sensor data over MQTT to Pi as a subscriber. |
| 5 | Management Protocol: CoAP | Introduction to RESTful model for constrained devices, CoAP message types (Confirmable, Non-confirmable), Resource discovery. | Raspberry Pi, ESP8266. Lab: Implementing a CoAP server on the Pi to allow the MCU to request device state/configuration. |
| 6 | Open Source Data Management | Introduction to open-source time-series databases (e.g., InfluxDB), Data schema design for edge data, Data ingestion from MQTT subscribers into the database. | Raspberry Pi (hosting DB). Lab: Setting up InfluxDB and writing sensor data from Week 4 via a Python script. |
| Phase 3: Advanced Edge AI and System Monitoring (Weeks 7-10) | |||
| Week | Topics | Key Concepts & Labs | Hardware Focus |
| — | — | — | — |
| 7 | Advanced AI: Introduction to NVIDIA Jetson | Jetson Nano architecture (GPU), CUDA, Deep Learning frameworks (PyTorch/TensorFlow) setup, Introduction to accelerated computing for AI at the Edge. | NVIDIA Jetson. Lab: Initial setup and running a basic GPU-accelerated “Hello World” example. |
| 8 | Edge Vision AI Agent Development | Computer Vision basics, Pre-trained models (e.g., MobileNet, YOLO), Model optimization for Jetson (e.g., using NVIDIA TensorRT). | NVIDIA Jetson with a Camera Module. Lab: Developing a simple object detection or image classification AI agent. |
| 9 | System Monitoring and Visualization | Introduction to open-source visualization tools (e.g., Grafana), Data source configuration (connecting to InfluxDB), Dashboard creation for real-time monitoring of agent state and sensor data. | Raspberry Pi (hosting Grafana/visualization layer). Lab: Creating dashboards to visualize sensor data and AI agent performance metrics (e.g., inference time). |
| 10 | Final Project: Integrated Edge AI System | Integration of all components: MCU sensors -> MQTT -> Database -> Jetson AI Agent (subscribing to events/data) -> Visualization. System robustness, security, and next steps in MLOps. | All Devices. Final Lab: A capstone project where the ESP8266 sends environmental data, the Jetson runs an AI task, and the whole system is monitored via Grafana on the Raspberry Pi. |
While this syllabus is ambitious in scope, experience has shown that software setup and installation consistently impedes course progress. Students often spend valuable lab time troubleshooting dependencies, version conflicts, and platform-specific configuration issues rather than engaging with the core learning objectives. To address this critical implementation challenge, I propose developing pre-configured system images for both the Raspberry Pi and NVIDIA Jetson platforms, with all required packages, dependencies, and frameworks properly installed and tested. This approach would allow students to focus on building and deploying AI agents rather than wrestling with infrastructure setup.
I am currently dedicating time to developing comprehensive software lists, installation procedures, and system images for this course. This work serves a dual purpose: it will streamline the educational experience while also supporting my broader research in creating an IoT management framework for art installations and Robot Art projects. The technical infrastructure developed for this course—particularly the edge AI agent architecture and deployment methodology—has direct applications in managing distributed sensor networks and intelligent systems for interactive art installations.
I am actively seeking collaborators to contribute to this initiative. Specifically, I welcome assistance in refining the curriculum content, developing and testing the deployment packages, creating laboratory exercises, and documenting installation procedures. If you have expertise in edge computing, AI deployment, IoT systems, or educational curriculum design, I invite you to join this effort to create a robust, accessible learning platform that can benefit both educational and creative technology communities.
I have created a GitHub repository Edge_AI_Co-op as the repository of documents and developed software.
Leave a comment