Watchful Blog

If you develop machine learning models, you probably spend much more time thinking about your code than about how you store the training data that you feed into that code