MetaFlow: Metadata-Driven AI Agent for ML Pipeline Designing

Summary

Build an AI agent that reads dataset metadata to architect, write, and execute a complete ML pipeline. • Automate the transition from a raw CSV to a runnable .ipynb notebook. • Reason through metadata (not raw rows) to ensure data privacy and statistical accuracy. • Provide a "Self-Correcting" feedback loop where the agent audits its own code before delivery.