Design a scalable logging and evaluation pipeline to monitor the performance of machine learning models in production at Meta Platforms.
