I wanted to check your bold claim of explaining and solving these ML problems, so I checked out the crypto prediction work:
In the introduction you state all the things that could affect the value of asset and then use none of them!
The ML model generation lacks any quality checks and no validation of the output. Since your predictions were for Jan 2022, how did you do? I suspect not great as the only features you used were er... the historical price!
The code is just a copy and paste from the intro of the docs and certainly no explanation. Feels like you just listed a lot of potential projects where you spend 10 mins on actual ML work.