Machine Learning Design Patterns is a very helpful catalog of problems that ML practitioners commonly face, and design patterns to solve them. Building machine learning systems comes with a suite of problems related to, and distinct from, typical systems and software engineering ones. Think for example of reproducibility, serving, feature storage and retrieval, and model training.
As the authors put it, having named a pattern saves us the added effort of its continual rediscovery. This book is great fun to keep nearby, to skim occasionally for a chapter related to a problem you have at hand. Reading any given pattern in this book usually reminds me of one more ML problems that would have been much easier if I’d had the pattern in my problem-solving vocabulary.