Introduction to Machine Learning Systems

An innovative textbook that uses a systems approach to teach students and practitioners how to engineer machine learning systems that are reliable, efficient, and scalable in real-world settings.

This groundbreaking textbook provides a comprehensive framework for understanding and engineering machine learning systems, emphasizing the systems perspective required to build effective AI solutions. Unlike resources that focus primarily on algorithms and model architectures, Introduction to Machine Learning Systems integrates engineering principles, system abstractions, and practical techniques to bridge the persistent gap between theoretical foundations and production. It covers the full lifecycle, including systems foundations, data pipelines, training and inference infrastructure, deployment, monitoring, benchmarking, security, privacy, and sustainability. The scope spans edge devices, embedded systems, and large-scale cloud platforms. Following a pedagogical progression that mirrors how expert engineers develop their skills, this learn-by-doing text equips readers to reason about machine learning system architectures and apply enduring engineering principles to build flexible, efficient, and robust machine learning systems.

• Provides end-to-end coverage of the machine learning systems lifecycle
• Emphasizes benchmarking, performance, and empirical rigor
• Offers rich pedagogy including learning objectives and self-check questions 
• Integrates with open-source tooling and real system case studies
• Based on the author’s popular Harvard course and class tested by thousands of students worldwide
• Features extensive supplemental resources including labs

| Vijay Janapa Reddi is the Gordon McKay Professor of Electrical Engineering at Harvard University, where his research focuses on the intersection of computer architecture, machine learning systems, and autonomous agents. He is Vice President and cofounder of MLCommons, a nonprofit organization dedicated to accelerating machine learning innovation, and created the Tiny Machine Learning edX series, a global MOOC that has trained over 100,000 learners.