Thanks for the warning! To elaborate, I have come to understand that code is just a part of the paper, seldom the beef. I basically create proofs of concepts, which I've told increase the impact of the paper (thus be fine to work on). I find that in general, it's not about how much or elegant code I write, but whether it allows me and my peers to learn new.

Industry labs are a good point, but I think I have to consider those later when my academic profile is stronger. Beyond Google X, do you know other options?

And I do have the burning question, which is the end-goal mentioned in the OP.

Thanks for the comment, I feel like there's a pressure for me to create a startup (market validated, employees and funding found) but I feel like academia is actually what I want. Your points solidify those thoughts.