Thread by @lak_gcp

Five months later, our ML patterns book is #3 in AI, behind only the top ML intro book and the top research one. Very grateful for the validation ... W/ @SRobTweets
amazon.com/Machine-Learni…
Like most authors, we keep hitting F5 to read the reviews 😁 My favorites 🧵👇

"When I was learning C++, I found the Gang of Four book "Design Patterns" accomplished a similar goal to help bridge the gap between academic knowledge and practical software engineering. Much like with the GoF book I suspect I may be re-reading parts of this book in the future"

"must-read for scientists and practitioners looking to apply machine learning theory to real life problems. I foresee this book becoming a classical of the discipline’s literature."

"This book provides an excellent reference point for me to cross-validate some of my empirical knowledge. It also has comprehensive discussions on many tradeoffs, which only seasoned ML practitioners will know."

"I found myself referencing the book for design advice almost daily and I am very much looking forward to implementing a variety of design patterns in the coming months."

"surprisingly this is the first of its kind to address common design patterns. Good ML design patterns hold their relevance over time much more than a framework or architecture might, so it's surprising that this book stands alone in this topic."

"sample code from TensorFlow and Keras, or BigQuery, which can be downloaded from GitHub.
Basically, the focus is on “why is it effective” rather than “how do you realize it”, so if you are using other machine learning frameworks such as PyTorch I think it is a very useful book."

The reason these are my favorite reviews? Because they match almost point-by-point our discussions when we wrote the book: we consciously modeled on the GoF book but wrote it for practitioners rather than researchers, captured tribal knowledge and design tradeoffs, and

focused on the reasoning rather than on the implementation. So that the book would be relevant even for those not using Tensorflow, Keras, or Google Cloud. But we added implementation to keep things concrete. This was a hard balance. Very heartwarming to see that readers noticed!