Data Platform
Create, clean, extend, and manage the datasets behind your verification and crowdsourcing programs.
The Data Platform is one of Crowdee's four ways to work with Crowdee, alongside the Verification Platform (buy a verified result), Crowd Platform (run your own crowd program), and AI Platform (finetune your own model). It's where the files behind those other offerings actually live: versioned datasets, a unified pipeline catalog (13 pipelines), and the primitives they all build on.
Upload files into a named, versioned dataset, run a pipeline against a version, and feed the result straight into a verification pipeline run or a crowdsourcing job's input data — without a manual re-export step. The catalog covers three kinds of pipeline:
- Language Technology (7 pipelines) — transcription, translation, OCR, language identification, and entity detection. Produce structured JSON output, not a new file.
- Cleaning & transformation —
clean-audio-silence(trims silence),convert-format(audio format conversion), andpii-redact-text(LLM-based PII redaction). Each produces a new derived dataset version. - Dataset management —
dedup-content-hash(flags duplicate files for review, in place),split-dataset-version(partitions a version into train/val/test), andcrowd-label-classify(has a crowd assign a label to every file, with per-file consensus).
Datasets
Named, versioned collections of files with full lineage — the raw material for cleaning, enrichment, and pipeline runs.
Pipelines
All 13 Data Platform pipelines: Language Technology, cleaning/transformation, and dataset management.
Where to Go Next
- Getting Started — authentication, organizations, and your first API call.
- API Reference — the REST API surface behind datasets and pipeline runs.
- Verification Platform — run a pipeline against a dataset version to get a verdict.
- Crowd Platform — seed a crowdsourcing job's input data from a dataset.
- AI Platform — finetune a small open-source model on a crowd-labeled dataset version.
How is this guide?