feat(rag): add fastembed/faiss-cpu/pypdf for retrieval layer

.gitignore

@@ -34,3 +34,10 @@ mlartifacts/
 .idea/
 .vscode/
 .DS_Store
+
+# RAG knowledge base (user-supplied PDFs/MD; not source-controlled)
+data/knowledge_base/*.pdf
+data/knowledge_base/*.PDF
+
+# RAG built artifacts
+data/processed/faiss_index/

requirements.txt

@@ -37,6 +37,11 @@ pytest==8.3.3
 pytest-cov==5.0.0
 httpx==0.27.2  # FastAPI test client
 
+# --- RAG (knowledge retrieval for agent feedback loop) ---
+fastembed==0.4.2          # ONNX-based embeddings, no torch dep
+faiss-cpu==1.8.0          # vector store
+pypdf==5.0.1              # PDF text extraction
+
 # --- Frontend (B2B dashboard) ---
 streamlit==1.39.0