I think it could be more valuable to find a problem you find interesting and see if there is a model in the literature that you can use / implement starting wit

I think it could be more valuable to find a problem you find interesting and see if there is a model in the literature that you can use / implement starting with the most general/available and specializing if necessary. Most of ML work is, as the article alludes to, collecting and managing data.