Building a Full RAG Pipeline with LangChain: From Document Loading to Multilingual Generation