finds.dev

DataWithBaraa/databricks_bootcamp_2026

End-to-end Data Lakehouse project built on Databricks, following the Medallion Architecture (Bronze, Silver, Gold). Covers real-world data engineering and analytics workflows using Spark, PySpark, SQL, Delta Lake, and Unity Catalog. Designed for learning, portfolio building, and job interviews.

View on GitHubHomepage ↗

★ 458Jupyter NotebookMITupdated Jan 2026

// the take

A bootcamp project (not a reusable tool) showing a bronze/silver/gold medallion lakehouse built on Databricks with PySpark, Delta Lake and Unity Catalog, using toy CRM/ERP CSVs for a fictional bike company. It's aimed at people learning data engineering for portfolio or interview purposes, built to accompany a paid course and YouTube series rather than to be used as-is.

The layer split is genuinely instructive: bronze ingestion, silver cleaning/type-casting, and gold dimensional modeling are each their own notebooks instead of one monolithic script, so you can see where responsibility actually shifts in a real pipeline. It uses real Databricks primitives (Delta Lake, Unity Catalog, Spark SQL) rather than faking them with plain Spark, so the patterns transfer to an actual Databricks workspace. The gold layer builds a proper dimensional model (fact_sales plus dim_customers/dim_products) instead of just dumping cleaned tables, which is the part most tutorials skip.

It's static example code with toy CSVs (a few thousand rows at most), so none of the Spark/Delta choices here get stress-tested — you won't learn anything about partitioning, skew, or cluster sizing from running this. There's no CI, no data quality checks, and no tests on the transformations, so 'correctness' is whatever the notebooks happen to output; nothing would catch a broken join. The author explicitly tells you not to use the repo directly — it's meant as an answer key after you've built the thing yourself from a Notion roadmap and paid/free course content, so a chunk of the real instructional value lives outside the repo. There's no orchestration beyond notebook-to-notebook calls (an 'orchestration' notebook, not a scheduler/DAG), so it won't show you how this would actually run unattended in production.

We read the README, the directory tree and a key source file in November 2026. Nobody pays to be listed. How we judge →

// at a glance

Stars
458
Forks
201
Language
Jupyter Notebook
License
MIT
Last commit
19 Jan 2026

// nearby

If databricks_bootcamp_2026 isn't quite it.

All repos →