A repository containing some projects I deployed in Azure Machine Learning, focusing on MLOps, Azure ML CLI and GitHub Actions. This also includes some modified versions of labs I completed for Azure ML SDK v2.
- Implemented different environments in Github with Azure User-Assigned Managed Idenitity
- Configured CI/CD using Github Actions including linting, unit testing, automatic deployment and endpoint set-up
- Upon PR to main, the training script is run/test automatically and if successful, a new production run is created upon approval. Once merged to main, an endpoint is updated/created and the model is re/deployed.
- Used Azure curated conda environement for Python 3.8 (a required fix as Azure curated environment for 3.7 crashes due to dependency resolution issues)
- An MLFLOW model (Logistic Regression) deployed on a real-time endpoint in Azure: