Warehouse

Unstructured Archive Ingest

Drop zip archives of PDFs, images, and documents into bronze storage for downstream structuring and ML pipelines.

Drop zip archives of PDFs, images, and documents into bronze storage for downstream structuring and ML pipelines.

Business problem

Legal, insurance, and field operations teams accumulate years of documents, images, and scans in shared drives. Moving that data into analytics or ML pipelines usually requires bespoke preprocessing jobs.

Heimdall solution

Lake unstructured ingest accepts zip archives into bronze storage with catalog metadata. Files stay organized by project or batch, ready for Lab exploration or ML feature extraction on gold publish.

Business impact

Archive batches become searchable catalog entries instead of orphaned folders. Downstream teams know exactly which bronze batch fed a model or report.

How it works

1

Upload zip archives

Drop multi-file archives into Lake bronze. Heimdall catalogs contents and preserves batch metadata for traceability.

2

Explore in Lab

Profile file counts, types, and sample metadata with Python before committing to a curation strategy.

3

Feed ML pipelines

Publish gold structured outputs and train classification or extraction models in Heimdall ML.

See Heimdall in action

Describe the document or image archives you need to ingest

Key benefits

Batch traceability

Know which archive version trained each model

No one-off scripts

Ingest through Lake instead of custom unzip jobs

Mixed media support

PDFs, images, and documents in one bronze batch

Catalog visibility

Browse archive contents before processing

ML-ready path

Connect to Heimdall ML after gold publish

Compliance friendly

Retain lineage from raw archive to model input

Heimdall products for this workflow

Ingest unstructured archives in Lake

Bring zip batches into bronze and catalog them for your team.