The data engineering team faces significant inefficiencies in manualy converting Excel files into SQL queries, particularly when dealing with bilinual or multilingual files. This labor-intensive process consumes over 40 h per week and is prone to errors, with an estimated 15–20% inaccuracy rate in query generation. To address these challenges, we developed an AI-powered virtual assistant that automates Excel-to-SQL conversion using Large Language Modls (LLMs), LangChain, and LangGraph. Our solution streamlines data extraction, SQL generation, and validation while integrating human-in-the-loop (HITL) feedback to ensure accuracy. By reducing manual effort and minimizing errors, this approach allows data engineers to focus on higher-value tasks, significantly improving productivity and reliability in data pipeline workflows.