We've been delving into the new Azure ML Designer and find it promising for initiating model training experiments. However, we're keen on incorporating MLOps to iteratively train models and deploy stable versions to higher environments.
We've checked the MLOps v2 documentation and have set up the starter pipelines in Azure DevOps as per the documentation. The starter repository includes a Python-based experiment deployed as a model, which is then pushed to a higher environment. This serves as an excellent example for code-driven models.
However, we're curious about best practices or guidelines for applying the same MLOps approach to models built with the Azure ML Designer pipelines. Any insights or recommendations would be greatly appreciated.
Thanks in advance!