Cue Validation in memory in Python

[suryaiyer95]

I want to apply a large number of validations at once. I want to do them in memory in python

A few questions:-

1. Is it possible to do cue validations in memory in python? I do not want to parse the row to a json and save and then validate files in cli. Also, I would not be able to use the Go API
2. Is it possible to translate the basic constraints into python objects using codegen? I tried to convert the cue to openApi json and then to python models. But it looks like the constraints were not generated.
3. Is there anything on the roadmap for codegen/in-memory support for python and other languages?
4. The below cue file cannot be translated to openapi. Any approaches on how we can have such constraints exported as code?

#schema: {
a : int
b : int &  < a
}

The alternate approach I see is to build a parser in python specific to my use case. But It feels like I am losing cue's main functionalities when I do that.

Is there any set of best practices for doing a large number of validations in a optimal manner?
Appreciate your help and suggestions!