As a fellow evangelist of the SQL + Pandas hybrid workflow, I’m a happy camper with Pandas’ built-in read_sql_query and to_sql.
Only big pain points are having to ship around boilerplate to construct SQLAlchemy create_engine URIs, and the performance limitations of SQLAlchemy’s inserts (if moving anything larger than a few gigs, it typically pays to ditch to_sql, and write a db-specific bulk insert process instead).
Only big pain points are having to ship around boilerplate to construct SQLAlchemy create_engine URIs, and the performance limitations of SQLAlchemy’s inserts (if moving anything larger than a few gigs, it typically pays to ditch to_sql, and write a db-specific bulk insert process instead).