And ML is a tiny part of the whole data science toolchain and process. Getting the data to the point where you can use it for something interesting is probably the hardest part.