Overview

This skill covers the application-layer patterns for running Qdrant inside a FastAPI backend — the glue between your async web framework and the vector store. It does NOT duplicate the official Qdrant client docs; instead it focuses on what changes when Qdrant is a dependency of a production FastAPI service serving multiple tenants.

What this skill covers (that the official qdrant/skills does not):

Lifespan-managed client injection so one QdrantClient lives across requests
Tenant isolation via payload filters (single collection, many tenants)
Batch upsert with idempotent UUID5 IDs — safe to re-run ingestion
Payload indexing strategy for filtered search at scale
Connection retry, error handling, and structured error responses
Docker Compose deployment alongside FastAPI + Redis + Postgres

When to use qdrant-fastapi-integration vs. the official qdrant/skills: