📊 AI-Powered Excel Interview Bot

An AI-driven Excel interview platform built with LangGraph, FastAPI, and Streamlit, designed to simulate a real-world adaptive interview. The system verifies candidates, asks dynamic Excel questions, evaluates responses, adapts difficulty, generates structured reports, and saves results into MongoDB for HR review.

---

🎯 Why This Project?

Hiring managers often struggle with scalable and fair Excel skill assessments. Manual interviews:

• Take time ⏳
• Are subjective 🤔
• Don’t scale across candidates 👥

This project solves the problem by building an autonomous AI-powered interviewer that is:

• 🔄 Adaptive → difficulty adjusts based on candidate performance.
• 📝 Structured → every question, answer, evaluation, and score is logged.
• 📊 Insightful → generates reports with strengths, weaknesses, and pass/fail decision.
• ⚡ Production-Ready → integrates directly with HR systems.

---

🚀 Features

• 🔑 User Verification (Email/Employee ID with retry & error handling)
• ❓ Dynamic Questioning (topic-based Excel interview questions)
• 🤖 AI Evaluation (LLM-based grading out of 10)
• 📈 Adaptive Difficulty (questions get harder/easier based on scores)
• 📂 Answer Logging (all Q/A/Evaluations stored in MongoDB)
• 📝 Structured Report Generation (strengths, weaknesses, final decision)
• ☁️ MongoDB Storage (audit-friendly logs and results for HR)
• 🖥️ Streamlit Frontend (chat-like UI for candidate interaction)
• ⚡ FastAPI Backend (async, modular, and production-ready)

---

🛠️ Tech Stack

• Backend: FastAPI + LangGraph
• Frontend: Streamlit (chat-based UI)
• Database: MongoDB (Async)
• AI/LLM: OpenAI GPT-4o-mini for Q/A generation & evaluation
• Orchestration: LangGraph (interrupts, adaptive flow, state management)

---

📂 Project Structure

Directory structure:
└── rawatrahul14-excel-interview-bot/
    ├── README.md
    ├── app.py
    ├── main.py
    ├── requirements.txt
    ├── setup.py
    ├── template.py
    └── src/
        └── interviewBot/
            ├── __init__.py
            ├── agent_state.py
            ├── details.py
            ├── graph.py
            ├── Agents/
            │   ├── __init__.py
            │   ├── adaptive.py
            │   ├── answers.py
            │   ├── error.py
            │   ├── evaluate.py
            │   ├── final.py
            │   ├── initSess.py
            │   ├── make_report.py
            │   ├── question_maker.py
            │   └── user_verification.py
            ├── components/
            │   ├── __init__.py
            │   ├── report_upload.py
            │   ├── topic.py
            │   └── verification.py
            ├── data/
            │   └── questions.yaml
            ├── routes/
            │   ├── __init__.py
            │   └── router.py
            ├── Schema/
            │   ├── __init__.py
            │   ├── model_output.py
            │   └── responses.py
            └── utils/
                ├── __init__.py
                ├── common.py
                └── ui_streamer.py

---

⚡ Installation & Setup

1️⃣ Clone the Repository:

git clone https://github.com/RawatRahul14/excel-interview-bot.git
cd excel-interview-bot

2️⃣ Create Virtual Environment:

python -m venv .venv
source .venv/bin/activate   # Mac/Linux
.venv\Scripts\activate      # Windows

3️⃣ Install Dependencies:

pip install -r requirements.txt

4️⃣ Configure Environment: Create a .env file in the root folder:

# === MongoDB Connection ===
MONGODB_URI = ""
DB_NAME = ""
COLLECTION_NAME = ""

# === Verification Collection ===
DB_NAME_VERIFICATION = ""
COLLECTION_NAME_VERIFICATION = ""

# === Upload Collection ===
DB_NAME_UPLOAD = ""
COLLECTION_NAME_UPLOAD = ""

# === OpenAI API Key ===
OPENAI_API_KEY = ""

5️⃣ Run Backend (FastAPI):

uvicorn backend.main:app --reload

6️⃣ Run Frontend (Streamlit):

streamlit run app.py

🖼️ Workflow Diagram

Steps:

• Candidate verification (email/ID)
• Dynamic Excel questions asked
• Candidate responds → evaluated by AI
• Difficulty adapts automatically
• All answers saved to MongoDB
• Structured report generated
• Final decision → Pass / Fail / Needs Review

---

📊 Example Output

MongoDB log document:

{
    "user_id": "rahulrawat272chd@gmail.com",
    "timestamp": "2025-09-16T17:09:36.817+00:00",
    "answers_log": [
        {
            "q": "Explain how to use Solver...",
            "ans": "I'd set up profit as objective...",
            "eval": "Good explanation with clear steps",
            "score": 8
        },
        {
            "q": "Differences between VLOOKUP, INDEX-MATCH, XLOOKUP...",
            "ans": "VLOOKUP is simple but limited...",
            "eval": "Good explanation with clear distinctions",
            "score": 8
        }
    ],
    "report": "Candidate has strong understanding...",
    "final_decision": "pass"
}