Online shopping platforms often struggle to deliver quick, relevant responses to customer questions about products, pricing, availability, and general policies, which can negatively impact engagement and sales conversion. To address this challenge, I developed ECommerceChatbot, an AI‑powered assistant that augments traditional FAQ support with context‑aware conversational capabilities. The objective was to create a chatbot that handles both general customer queries and structured product inquiries within a unified interface, demonstrating practical application of vector retrieval and language model integration to improve user experience.
I architected an AI chatbot system that leverages semantic search and vector retrieval via ChromaDB for FAQ handling, paired with an LLM backend for natural language generation. The solution ingests and vectorizes FAQ content, enabling fast semantic matches for user questions, while product and pricing details are stored in a SQLite database for structured lookups. A Streamlit frontend provides an interactive chat interface where users input queries and receive dynamic, context‑aware responses. Under the hood, a semantic routing layer classifies queries to either the FAQ vector store or the product database, and results are then refined with model‑assisted natural language output.
The resulting prototype significantly enhances customer engagement workflows by delivering accurate, conversational responses across a range of e‑commerce query types—both general FAQs and detailed product information—without manual scripting for each scenario. The project highlights applied expertise in semantic retrieval, LLM integration, chat UI design, and full‑stack prototype deployment, all of which align closely with roles in machine learning engineering, NLP product development, and AI‑enabled customer experience solutions. The repository’s clean structure, documentation, and demonstration ready Streamlit UI make it a strong reference for future production deployment and extension.