Embedding API

Last updated: Mar 10, 2026 19:00 UTC

LIVELLO 2 service for text embeddings (nomic-embed-text-v1.5, 768d) and optional Qdrant storage.

Location

• Service: services/api_embedding/

Base URL / Port

Default port is 8010 (env: PORT).

Endpoints (implemented)

Defined in services/api_embedding/api/routes.py:

• GET /health
• POST /v1/embeddings/create — single text embedding
• POST /v1/embeddings/batch — batch embeddings (max 100)
• POST /v1/sync/postgres_to_qdrant — sync placeholder (currently returns zeros)
• GET /v1/stats

Metrics endpoint (Prometheus) is defined in services/api_embedding/main.py:

• GET /metrics

Request/Response schemas

Pydantic models live in services/api_embedding/schemas.py.

Key request types:

• EmbeddingRequest (text, model, store_in_qdrant, collection_name)
• BatchEmbeddingRequest (texts[], model, store_in_qdrant, collection_name)
• SyncRequest (limit, offset)

Service config env vars

Loaded in services/api_embedding/config.py:

• PORT
• LOG_LEVEL
• EMBEDDING_MODEL
• QDRANT_COLLECTION