Agentic Retrieval-Augmented Generation system for product search enhancements

This project presents a hybrid search solution designed to enhance information retrieval in the automotive spare parts industry, particularly for handling bulk product requests from customers. At its core is a Retrieval-Augmented Generation (RAG) architecture that combines keyword, sparse, and dense search methods to improve the relevance and accuracy of search results across large inventory databases. An agent-based RAG system is implemented to process bulk queries efficiently, with a Large Language Model (LLM) agent responsible for validating and approving the retrieved results for each spare part. A human-in-the-loop evaluation mechanism is included to support ongoing fine-tuning and improve system performance over time. To improve user interaction, a chatbot system is developed, allowing fast, conversational access to spare part and vehicle information. The entire solution is deployed through a user-friendly web application, enabling seamless and efficient use in business environments.