Machine Learning Pipeline Architecture Diagram

The prompt behind this diagram

Create a machine learning pipeline architecture diagram showing: raw data sources, a data ingestion service, a data validation and cleaning stage, a feature engineering module, a feature store, a model training service, a model registry, a model evaluation stage, a deployment/serving layer, a monitoring and drift detection component, and a feedback loop back into data ingestion. Show the sequential flow with arrows and include a batch/offline path and a real-time inference path branching from the feature store.

Paste your own description (or Terraform / docker-compose / SQL schema) into draft1 and get a diagram like this for your exact system.