How to choose a language and library for building a machine learning system?

Question

Machine learning

Hi all.
I want to try to write a small system with a text classifier. This is not production, but I want to do something similar to the truth. I'm interested in the principles of how this is done in real systems. I wrote similar things in python, but, as I understand it, prototypes are written in python, C ++ or java are already used in production with their own libraries like Spark MLLib or Weka (I don’t consider Amazon, because it’s paid).
I'm interested in learning at least in general terms how real machine learning systems work. And ideally, with a description of the languages ​​​​and technologies used.
I would be grateful for any thoughts on this topic or links or titles of books where you can read.
Thank you.