I’m in a similar situation as several of my third year PhD student friends–we’ve enjoyed doing research and have had various levels of success so far, enough to be able to (and want to) hopefully finish in the next two years, but we all realize that we’ll be doing industry roles and that it would be tough with our publication records to get into one of the top research groups that publishes pure research.

I’ve had some research-y internships in industry but I think a “research engineer”-type role would be best for me, (and it’s also much more common than research scientist positions in areas I’m interested in such as autonomy and robotics). However, those roles generally ask for someone who is equally well-versed in statistical machine learning theory, deep learning and Python (check) and software development in C/C++ (something I need to work on). Research code is almost exclusively in Python (aside from the ocassional CUDA kernel for a novel layer) so essentially no one I know does development in C++ aside from side projects.

How do you maintain your software chops while focusing on developing novel methods and writing research code in Python?