ML-Enabled Systems Model Deployment and Monitoring: Status Quo and Problems provides a comprehensive survey of the practices and difficulties faced during the ML model deployment and monitoring phases. Highlights include:

• Limited adoption of MLOps principles during the deployment phase.
• Challenges in architecting a suitable production environment and integrating with legacy applications.
• Many models in production lack active monitoring, and the selection of monitoring metrics is a noted issue.

These findings underscore the significance of streamlined deployment and rigorous monitoring practices in ML-driven systems, pinpointing areas for improvement and the need for adopting a more standardized and effective approach to MLOps.