30-Day Data Engineering Fundamentals Plan

This plan compresses the first stages of the Data Engineer roadmap into four weeks. Each week has one outcome. Do not move on until you can meet it; slipping a week is better than skipping understanding.

How to use it

• Block the hours in your calendar before the week starts.
• Write your answers and code in a Git repository so you can show your work later.
• At the end of each week, explain that week’s outcome out loud in two minutes. If you cannot, repeat the hardest part.

The plan

1. 01

   Stage 1: Week 1: SQL that answers questions

   Joins, aggregation and row counts. Write 20 or more queries against a sample dataset.

   Typical effort

   8–12 hours

   Outcome

   You can predict the row count of a join before running it.

   • SQL Joins Explained for Data Engineers
2. 02

   Stage 2: Week 2: Window functions

   Ranking, top N per group, LAG/LEAD and running totals.

   Typical effort

   8–12 hours

   Outcome

   You can solve top-N-per-group and deduplication with window functions.

   • SQL Window Functions: PARTITION BY, ORDER BY and Frames
   • How do window functions differ from GROUP BY?
3. 03

   Stage 3: Week 3: Model the data

   Facts, dimensions, grain and a star schema for one business process.

   Typical effort

   8–12 hours

   Outcome

   You have a written star schema with a stated grain.

   • Star, Snowflake and Galaxy Schemas and the Grain
   • ETL vs ELT: Choosing the Right Approach
4. 04

   Stage 4: Week 4: Build a rerunnable load

   Python loader with an upsert, a transaction, logging and tests.

   Typical effort

   10–14 hours

   Outcome

   You have a loader whose test proves running it twice changes nothing.

   • Build an Idempotent CSV Loader in Python
   • CSV to Data Warehouse Pipeline