Recently, I completed booked called AI Engineering by Chip Huyen. I noticed that there were part of the day where it was a struggle to read since a lot of material was 'something people know when they use AI tool' and some chapter definitely worth a read and got me very engage.

Because of those early chapters being a struggle to read, the book took way longer than it should to complete it. So this got me thinking if there is a way to 'crowd source' material from long text book to be 'must read' and other sections can be skipped due to overlap or being too basic.

So my question is for folks who recently read a book or material that really was an eye opener or material they got a lot of ROI on knowledge.

I'll start, it would had been great if someone told me must read chapter are the following from AI Engineering:

1. Finetuning (Chap 7)

2. Inference Optimization (Chap 8)

3. Dataset Engineering (Chap 9)

Rest I feel is very common knowledge and can be attained by briefly trying to build a RAG application.