As more and more ML models move into production, the sustainable quality of the data flowing into the models becomes crucial. We make sure to detect variations in data quality early on and take automated steps to correct obvious mistakes.