Ad-Hoc Transforms
Run Python transforms on Lake tables and save results as new silver datasets — without overwriting source tables.
Run Python transforms on Lake tables and save results as new silver datasets — without overwriting source tables.
Business problem
Business teams need one-off transforms — regional rollups, cleaned categories, joined enrichment — but engineering backlogs delay every request.
Heimdall solution
Lab notebooks let analysts run pandas transforms on Lake tables and save outputs as new silver datasets. Source bronze and silver tables stay untouched; every save creates a new catalog entry.
Business impact
Self-serve transforms ship in hours. Data engineering focuses on governance and gold publish instead of every ad-hoc request.
How it works
Load source tables
Reference bronze or silver by name in a Lab cell.
Transform in Python
Apply filters, groupbys, and joins with familiar pandas code.
Save to silver
Persist results as a new silver dataset — originals are never overwritten.
See Heimdall in action
Tell us about the transforms your team needs
Key benefits
Self-serve transforms
Analysts ship logic without ticket queues
Non-destructive saves
New silver tables; sources stay intact
Reproducible notebooks
Re-run when upstream data refreshes
Catalog integration
Saved tables appear immediately in Lake
Gold-ready path
Promote validated silver to gold when stable
Isolated workers
Sandboxed execution until explicit save
Related use cases
More scenarios on the use cases hub.
Multi-Source Consolidation
Unify CSV exports, Postgres tables, and warehouse snapshots into one bronze catalog — no ETL scripts to maintain.
Unstructured Archive Ingest
Drop zip archives of PDFs, images, and documents into bronze storage for downstream structuring and ML pipelines.
Gold Feature Publishing
Curate silver views, publish gold datasets, and connect them to ML and Forecast training jobs.