This demo contains 3 folders :

• duckdb-pypi : notebooks and sql to gives intro to DuckDB and play around the pypi dataset on MotherDuck
• md-dbt : same queries used from the above, to give an highlight on how duckdb/md and dbt works together
• md-evidence : for local snappy dashboard

You can find the videos of the demo here

See demo script on Notion

Requirements :

• VSCode & devcontainer extension
• Docker desktop
• A MotherDuck account & service token
• Copy queries that are in ./duckdb-pypi/motherduck_ui_queries.sql in the MotherDuck UI

Use the devcontainer feature from VSCode to load the container that will have all requirements :

• Python 3.11
• dbt/duckdb packages
• NodeJS (for Evidence dashboard)
• DuckDB CLI
• A parquet dataset downloaded from AWS S3 (in the dataset folder) of ~900MB

Open duckdb_intro.ipynb in VSCode and go through the cells (intro duckdb, query Pypi data and connect to MotherDuck Cloud).

You can use the two make commands

• make dbt-local : will run dbt/duckdb locally based on the parquet dataset present in ./dataset and output CSVs in md-dbt/output
• make dbt-md : will run dbt/duckdb locally and push the data to MotherDuck under the database my_db. You need to export first your MOTHERDUCK_TOKEN before running this!

You can use the make commands :

• make evidence-install: will install node packages
• make evidence-run: will run a local server, dashboard will be accessible typically on port 3001.