Two papers address the critical phases of model deployment and monitoring in machine learning:

• ML-Enabled Systems Model Deployment: This study investigates current practices and problems, finding that model deployment typically involves deploying models as separate services with limited MLOps principles, while monitoring is not universally employed.
• A framework for monitoring and retraining language models considers post-deployment activities essential in many applications. It discusses continuous monitoring, identification of concept drift, and data distribution changes.

The two studies shed light on the practical aspects of deploying and maintaining ML models, stressing the importance of an end-to-end approach that enhances the models’ real-world efficacy and longevity.