RAG-02 | ChromaDB Fallback

[Manish281005]

Task: RAG-02 | ChromaDB Fallback

Estimated Hours: 2h | Owner: MK | Priority: P1 | Status: Completed

Definition:
Implement dual-index abstraction: VectorIndex base class with two implementations (Azure AI Search for production, ChromaDB for local dev). Allows zero-cost demos and offline development without cloud keys.

Acceptance Criteria:

• VectorIndex abstract base class with index(), search() methods
• AzureAISearchIndex implementation (currently acts as a proxy wrapping the local index)
• ChromaDBIndex implementation (local persistence: ./data/chroma/index.json)
• Both use same embedding model (deterministic character-based stub) for consistency
• Factory pattern: select via USE_CHROMA=true env var
• ChromaDB setup requires zero external credentials