A pinned, multi-stage, non-root JupyterLab 4 data science image (26 exact package versions) plus a compose stack with Postgres 16 and a headless papermill batch runner, where every setting is an environment variable the buyer owns and the notebook token is never blanked.
docker compose up -d
Dockerfilerequirements.lock.txtscripts/entrypoint.shscripts/healthcheck.shconfig/jupyter_server_config.pyconfig/matplotlibrcjupytext.tomlcompose.yaml.env.exampledb/init/01-analytics.sqlnotebooks/00-start-here.ipynbMakefile.dockerignoreREADME.mdThis page is the working piece. The full pack has everything below.
14 files, 26 Python packages and 2 image tags pinned to exact versions, zero `latest`: 3 commands from an empty folder to JupyterLab reading your own CSVs, writing to Postgres 16 and saving
Measured on this shelf right now: a comparable ready-to-run data science image is being sold at 19 USD. We deliberately quote no analyst-hour or cloud-notebook figure, because we have no measured one — every number in this listing
Send it to me