Chain collections to build multi-stage processing workflows
source.type
to COLLECTION
and reference the upstream collection_id
video_scenes
(BUCKET → scene splitting) → scene_analytics
(COLLECTION → face object detection) → scene_enriched
(COLLECTION → taxonomy enrichment)docs_raw
(BUCKET → OCR and PDF parse) → docs_nlp
(COLLECTION → embeddings and entities) → docs_topics
(COLLECTION → topic tagging)images_raw
(BUCKET → EXIF and thumbnail) → images_semantic
(COLLECTION → CLIP embeddings) → images_moderation
(COLLECTION → safety labels)*_v1
, *_v2
collections side by side for safe rolloutsproducts_raw_v1
)