There are super-complicated SAT solvers certainly, but MiniSAT has (in my opinion) reasonably easy to understand code. It's not very long either. Have you tried reading the code to Tensorflow?
Also, in my personal experience, SAT is currently applicable to more problems than Neural Networks (although that is changing as Neural Networks get applied to more things). You can use SAT solvers to solve any problem in NP (well, other things as well but it rapidly gets painful). That's a lot of problems!
Also, in my personal experience, SAT is currently applicable to more problems than Neural Networks (although that is changing as Neural Networks get applied to more things). You can use SAT solvers to solve any problem in NP (well, other things as well but it rapidly gets painful). That's a lot of problems!