Quickstart

You need Docker with Compose, and the datakumi files. Five steps take you to a published dataset.

1. Start datakumi.

cp .env.example .env          # set MDM_SEED_DEMO=true for the demo, or MDM_ADMIN_EMAIL and MDM_ADMIN_PASSWORD for a clean start
docker compose up -d --build  # the app is at http://localhost:8000

2. Sign in at http://localhost:8000. With a clean start, use the admin you set in .env. You are asked to choose a new password. With the demo, sign in as one of the demo accounts.

3. Create a dataset from a spreadsheet. Upload an Excel or CSV file. datakumi finds the header row, suggests the types, the key and the required fields, and shows them for you to review before it creates the dataset.

4. Edit, then publish. Change a few values in the grid. Your edits are saved as a private draft. Choose Review & publish to see a summary of what changed and any rule problems. If the dataset needs approval, the change goes to its approvers first.

5. Read it from code. Every published version is a Delta table in your storage. The path is shown on the dataset's Settings tab.

import polars as pl

path = "s3://lake/commercial/product_master/products"
current = pl.read_delta(path)
release = pl.read_delta(path, version=1)  # any earlier version