ICICLE AI Vector Service

FastAPI + Qdrant vector storage and retrieval service for the ICICLE AI Tapis tenant. Clients provide their own pre-computed embeddings — the service handles storage, filtered search, and reranking.

API reference

This component exposes an HTTP API — see its API documentation on this site.

Every request is authenticated with a Tapis access token, and user_id is taken from the token's tapis/username claim — never from the request body. Each user's collections are separate Qdrant collections, so users cannot see or affect each other's data, and each can choose their own embedding model and vector dimension.

References

• Qdrant Documentation
• FastAPI Documentation
• Tapis Project
• Diataxis Framework

Acknowledgements

National Science Foundation (NSF) funded AI institute for Intelligent Cyberinfrastructure with Computational Learning in the Environment (ICICLE) (OAC 2112606)

Issue Reporting

Please report issues via GitHub Issues. Include steps to reproduce, expected behavior, and any relevant logs.