Version Everything: The ML Practice a Lot of Projects Skip

A long time ago, I was working on a model and doing what most people do early on: iterating fast, making transformations directly on the dataset to try to improve results, retraining, checking metrics