"The Hundred-Page Machine Learning Book" will teach you what ML is. It can be a treated as a light weight replacement for say an Andrew NG course. You wont learn anything practical / code something in PyTorch or Tensorflow but you ll understand what happens under the hood in these framework

"Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems" is an amazing introduction to implementing all ML ideas. I think there is a new PyTorch version. The accompanying notebook will get you started to a point where you can do some hobby projects.

This book surprisingly seems to fill an interesting gap explaining about how these ML systems are used in real life in large scale. I work for 1 of the FAANG companies & I can say that every chapter here would correspond to bread & butter of a team responsible for maintaining a large ML system say Recommendation / Fraud detection. The target audience would be someone who is interested to learn how to put a large end-end ML system together.

I would be very excited if there are practical examples on how to use this with MLFlow / KubeFlow / Sagemaker. Really excited to read this

edit : typos