How do you store your ML datasets?

How do you store you ML datasets? I'm looking for something like an artifactory but optimized for ML datasets, where I can:

1) store sparse matrices (labels + features + comments) in an efficient way

2) push/pull a dataset quickly from a language or using a cli

3) (bonus) visualize the distribution of a particular feature and it's correlation with the label

Do you know any open source project providing these functionalities ?