How to implement GPU parallelization in Python having a Fortran subroutine?

Question

I am trying to speed up a program I wrote in Python by using GPUs, but I have almost zero experience with them (apart from some tutorials). I looked into Numba and CUDA, but I h