deepshikhar23/ai-ms-excel-interviewer

🤖 Excel Pro-Sim: AI-Powered Interviewer

A fully autonomous, multimodal AI agent that conducts practical mock interviews for Microsoft Excel skills.

About The Project

This project is an AI-powered assessment tool designed to automate the technical screening of candidates for Excel proficiency. Instead of a simple Q&A, this agent presents candidates with real-world tasks, provides them with unique datasets, and evaluates their solutions by analyzing both their written explanation and visual proof in the form of screenshots.

Key Features

• —🤖 Multi-Agent System: Built with LangGraph, the application uses a robust, stateful agentic architecture to manage the interview flow.
• —👁️ Multimodal Evaluation: Leverages Google's Gemini Vision model to analyze screenshots of the user's work (formulas, Pivot Tables, etc.) providing undeniable proof of skill.
• —⚙️ Dynamic Task Generation: Each interview session generates unique CSV datasets to ensure a novel challenge for every user.
• —🗣️ Interactive Experience: A web-based, conversational interface built with Streamlit.
• —🧠 Adaptive Logic: The graph is designed to be easily scalable with an adaptive engine that can adjust question difficulty in the future.

Tech Stack

• —Orchestration: LangGraph
• —LLM: Google Gemini 2.5 Flash Lite
• —Frontend: Streamlit
• —Core Libraries: LangChain, Pandas
• —Deployment: Hugging Face Spaces