I love initiatives like this, but IMO if you're doing a self-study "degree" aimed at practical AI knowledge, I would: * Drop the compilers and database courses

I love initiatives like this, but IMO if you're doing a self-study "degree" aimed at practical AI knowledge, I would: * Drop the compilers and database courses. * Add an intro to statistics course. * Pick one domain, since you'll need deep, proven domain expertise in one area to get hired without a real degree. So doing both NLP and CV, for instance, is probably a bad choice if this is your endgame. * If you choose CV, add a traditional image processing course. I don't recommend to rely on whatever happens to be included in the DL for CV course. You might also want to add a good general purpose DSP course. * Substitute exams for projects. Exams are adversarial so they don't make sense to give to yourself. You'll learn more more doing a project than you would cramming for a fake exam. * Try to finish the whole thing in 1.5 years, not 4. Spending 4 years on a practical curriculum that doesn't yield an actual degree is a huge waste of time. Try to get the theory out the way in a year and a half so you can jump into real life projects/research. With dedication and without a rigid university schedule that forces slow progress, this is completely doable.