Knowledge layer
Enterprise RAG Platform
Production RAG is a governed intelligence system — not a vector DB wrapper.
Authorization before ranking, hybrid retrieval, cross-encoder rerank, decline-to-answer, citation traceability, AegisAI HITL bridge, and Langfuse trace export.
In one sentence
Access-aware hybrid RAG with ingest, citations, and optional HITL gates.
Decision
Authorization before ranking — vector DB is implementation; access control is architecture.
Measured signal
golden-eval-registry CI · adversarial suite · JWT Principal when PRODUCTION_STRICT=1.
Honest limitation
Default live demo is Demo mode (body Principal) with sticky banner; Strict/JWT is the Principal review path (ADR-0006). Free-tier corpus re-ingests after cold start.
- Access-aware filtering before semantic ranking
- Hybrid lexical + semantic retrieval with cross-encoder rerank
- Decline-to-answer when retrieval confidence is low
- JWT-verified Principal under PRODUCTION_STRICT · adversarial golden suite
Video walkthrough — 8 min (planned)
Enterprise RAG: access-aware hybrid retrieval
Upload a document, compare three retrieval strategies, and inspect citations with authorization-before-ranking.
Recording in progress. Subscribe on YouTube to get notified when this walkthrough publishes.
Subscribe @venkat-aiArchitecture diagram
FastAPI · Docker · Vercel · Render · Qdrant (optional)