How do you prioritize tasks when working on multiple machine learning models with tight deadlines?