RetailAgent - Agentic Retail Analytics System

Agentic Retail Analytics System

  • Developed an agentic system enabling natural language queries over Statistics Canada retail datasets using LLM-driven reasoning.
  • Integrated SQLite for structured data storage, ChromaDB for vector retrieval, and Gemini for natural language understanding and response generation.
  • Achieved 0.90+ faithfulness and 4.6s average latency, validated using DeepEval benchmarking framework.
  • Deployed as an interactive Streamlit application for real-time analytics and user interaction.

Tech Stack

Python Streamlit SQLite ChromaDB Gemini DeepEval

User Interface

Retail Agent