You are building a machine learning model for a digital creative platform to predict whether a newly published asset will receive strong user engagement shortly after launch. The goal is to help rank and surface promising assets earlier, and you need to choose an approach that can be trained, evaluated, and maintained in production using standard machine learning frameworks.
How would you approach this problem, including which machine learning framework and model family you would use, how you would prepare the data, and how you would evaluate whether the model is good enough to ship?