The project will design and implement workflows covering the entire data management lifecycle for executing machine learning models: data collection, cleaning and processing for defining a training set, training/validation/testing of the machine learning model, as well as deploying the model for use. For these workflows, methods for automatic detection of quantitative and qualitative characteristics of the models affecting their performance will be developed, as well as methods for profiling data sets to automatically detect statistical characteristics causing issues in model performance. The methods will be evaluated in pilot scenarios of healthcare and energy services.