Legality of a basic open source CUDA backend

Hi, I’m thinking about developing a very basic and tiny subset of nvcc, specifically, I’m thinking about making a custom CUDA IR backend (IR → PTX assembly) - what’s the legality of this? I personally think its fine, since, as far as I know, LLVM has its own PTX emitter, but I’m not 100% sure.
I’m asking this because I might want to expand this project into a bachelors thesis or something similar in the future, so I’d like to be sure that its alright academically and legally.