Taxi Data Prediction Dashboard — Snowflake + Snowpark

ETL pipelines and ML transformations ran outside the warehouse, duplicating data movement

No single source of truth for both analytics and ML feature engineering

Snowpark Python UDFs run ML transformations directly inside Snowflake — no data egress

Jupyter notebooks handle EDA; SQL scripts define transformations; Streamlit Cloud hosts dashboard

Zero data movement between warehouse and ML layer

End-to-end pipeline in one platform: ingest → transform → model → visualise

Streamlit dashboard connected live to Snowflake for real-time analytics