How we build Feature Store

For more than five years, Big Data MTS has grown from a small internal startup into an independent center with more than three hundred employees and dozens of product teams. ML developments go through the stage of maturation from a rebellious rnd to a mature production. And caught a bunch of childhood illnesses: the team couldn't understand which features were already collected, where they are, how many models there are, what data they consume. There was no united means of supplying data. It turned out that these pains are universal for many companies, so MTS decided to look for possible solutions on the market.