Month 3 Capstone Project: Enterprise Hybrid RAG Engine

Blueprint Alignment: Project 3 of the 6-Month Staff AI Engineering Curriculum (ai-engineering-6-projects-blueprint.md) Production Capabilities:

1. Structure-Aware & Contextual Ingestion (Week 9): Parses Markdown headers/tables, builds [Context: doc=... | section=...] enriched chunks, and caches embeddings via SHA-256 content hashing.
2. Security & PII Shield (Week 12): Scrubs PII ([REDACTED_EMAIL], [REDACTED_SSN], [REDACTED_API_KEY]) and quarantines Indirect Prompt Injection payloads inside untrusted documents.
3. Two-Stage Hybrid Retrieval (Weeks 10 & 11): Combines SQ8 Quantized RBAC-Pre-Filtered Vector Search with Okapi BM25 via Reciprocal Rank Fusion (RRF, $k=60$), followed by Cross-Encoder Reranking and Lost-in-the-Middle Edge Packing.
4. Corrective RAG (CRAG) + Verbatim Citation Verification + Tenant-Isolated Semantic Cache (Week 12): Evaluates retrieval confidence before generation, verifies every cited quote deterministically, and caches verified responses per (tenant_id, role, corpus_version).