know what the business needs -> set performance metric -> get the data -> EDA to understand the data -> split data -> preprocessing for missing values, outliers, imbanace, etc -> modeling validation and parameter tuning -> A/B test the fnial model before fully deployment