Skip to main content

Qdrant

Qdrant is the vector database service for retrieval and embedding-backed knowledge pipelines.

Responsibilities​

  • Store vectors and payload metadata.
  • Support collection-level retrieval for RAG workflows.
  • Provide health endpoints and a dashboard where enabled.

Operational Notes​

  • Keep dashboard exposure restricted.
  • Validate collection health and vector counts after ingestion jobs.
  • Do not mix public and internal indexes unless explicitly designed and access controlled.
  • Use metadata filters to separate project, visibility, and source categories.

Related runbook: Qdrant Healthcheck.