dbt-duckdb just shipped 1.3.1 with a couple of very cool bits that are relevant when using it with meltano. • python models can be run inline. Since you are already running duckdb in-process, the extended python to also running on top of it (wherever your compute is, not limited to the cloud functions of your data warehouse). • first class support for reading/writing parquet for improved interoperability

Nice! https://github.com/jwills/dbt-duckdb#python-support