Pipeline Builder
Discover the no-code Pipeline Builder tool for creating transforms visually with a drag-and-drop interface.
1 · No-Code Transforms
Pipeline Builder lets you build transforms without writing code. You drag input datasets onto a canvas, add transformation nodes (filter, join, aggregate, pivot, derive column), and connect them to an output dataset.
Under the hood, Pipeline Builder generates the same Spark execution plan as a Python or SQL transform — so there is no performance trade-off.
2 · Common Operations
Pipeline Builder supports:
- Filter — keep rows matching a condition.
- Join — combine datasets on shared keys.
- Aggregate — group and summarise (count, sum, avg).
- Derive Column — add computed columns with expressions.
- Pivot / Unpivot — reshape wide to long or vice versa.
- Union — stack datasets vertically.
Each node shows a live data preview so you can verify results at every step.
3 · When to Use Pipeline Builder vs. Code
Pipeline Builder shines for straightforward ETL: filtering, joining, and aggregating. For complex logic — custom UDFs, machine learning, recursive algorithms — Code Repositories give you full flexibility.
Many teams use Pipeline Builder for 80% of their transforms and drop into code for the remaining 20%. Both approaches produce standard Foundry datasets.
✅ Key Takeaways
- Pipeline Builder lets you build transforms visually — no code required.
- It generates the same Spark execution plan as Python or SQL transforms.
- Live previews at every node make debugging straightforward.
- Use Pipeline Builder for standard ETL; use code for complex logic.