For GP, you can take a look at https://www.youtube.com/watch?v=4vGiHC35j9s

I think I learned it from this class like 8 years ago. GP is a very tricky topic and it is either explained too simply or too hard. I expect everyone will struggle with that at the beginning.

For Bayesian linear regression, it is not too hard. If you know the properties of normal distribution, you would know the details already. A basic college-level class in ML should cover it. Stanford, UBC, etc... have good classes on ML algorithms.

I am not sure I am much help here. The answer always depends on the background of the person asking the question. But in practice, there is no single book. And there is a major gap for people with engineering-based training rather than mathematical training. So Google a lot. Read a lot. Taking classes from major universities. That is how I start.