Built a RAG customer care chatbot for telecom support answering questions on connectivity, billing, SIM issues, and roaming, generating responses with Qwen3-32B served via Groq. Designed a merged-retriever architecture pulling top-k context from three ChromaDB collections — FAQ entries, resolved support tickets, and PDF guide chunks — for grounded, multi-source answers. Used sentence-transformers/all-MiniLM-L6-v2 for local embeddings and LangChain to orchestrate the retrieval-to-generation chain; shipped both a Streamlit web UI and a CLI.