HarvardX: Fundamentals of TinyML course Syllabus

This course provides a comprehensive introduction to TinyML, the emerging field of machine learning on ultra-low-power devices. Over approximately 8 weeks, learners will explore the foundations of edge AI, understand the technical constraints of microcontrollers, and learn how machine learning models are optimized and deployed in resource-constrained environments. With a conceptual focus, this course requires 4–6 hours per week, making it ideal for beginners seeking foundational knowledge in embedded machine learning.

Module 1: Introduction to TinyML and Edge AI

Estimated time: 6 hours

• Define TinyML and its distinction from traditional cloud-based ML
• Understand the importance of edge intelligence for latency, privacy, and power efficiency
• Explore real-world use cases of TinyML
• Identify key drivers in the adoption of TinyML

Module 2: Machine Learning for Resource-Constrained Devices

Estimated time: 8 hours

• Learn the memory, compute, and energy limitations of microcontrollers
• Understand how ML models are adapted for embedded systems
• Explore lightweight neural network architectures
• Study feature extraction techniques for efficient inference

Module 3: Model Optimization and Deployment

Estimated time: 8 hours

• Learn about model quantization and its impact on size and performance
• Understand pruning and model compression techniques
• Explore methods for reducing model footprint
• Conceptualize the TinyML deployment lifecycle

Module 4: TinyML Applications and Future Directions

Estimated time: 6 hours

• Explore applications in speech recognition and gesture detection
• Study sensor analytics in IoT and wearable devices
• Understand the role of TinyML in healthcare and smart manufacturing

Module 5: The Future of AI at the Edge

Estimated time: 4 hours

• Examine the evolving landscape of edge AI
• Understand how TinyML integrates with broader AI and IoT ecosystems
• Identify future trends and career-relevant opportunities in embedded ML

Module 6: Final Project

Estimated time: 6 hours

• Design a conceptual TinyML application
• Outline the workflow from data collection to deployment
• Present a use case with justifications for edge-based inference

Prerequisites

• Basic understanding of machine learning concepts
• Familiarity with Python programming (helpful but not required)
• Interest in AI, IoT, or embedded systems

What You'll Be Able to Do After

• Explain what TinyML is and how it enables AI on low-power devices
• Identify constraints in deploying ML models on microcontrollers
• Describe key optimization techniques like quantization and pruning
• Outline the TinyML workflow from data to deployment
• Recognize real-world applications in healthcare, wearables, and smart devices