sureshmrd/Agentic-AI-DataEngineering-and-Operations-Platform ? reverse-engineered prompt

Build me an end to end agentic AI data engineering and operations platform for the Olist e commerce dataset.

I want it to run monthly batch processing, take each batch through validation, transformation, and loading into MySQL, then refresh the analytics tables and track batch metadata and watermarks. It should use Airflow to manage the pipeline, with a manual trigger flow for historical batches and a deterministic notification step when a run finishes or fails.

On top of that, add a simple agent based interface where a user can ask for business insights, pipeline status, metric definitions, or help with failures, and the system routes the request to the right specialist. The business answers should use safe read only SQL, the troubleshooting agent should check Airflow logs and execution state, and the knowledge agent should use RAG over the project docs and schema info.

Please make it feel like a real working product, with a FastAPI backend, clear separation between the agents and tools, and tests for the main flows. Look up current docs online if you need to.