Supercharge Your RAG with DeepSeek's Reasoning Model