Minterviewer - Chrome Extension

Tutor and Mock Interview Practice

Features

• Chrome Extension

  • Clean chat-based interview interface
  • Uses random Leetcode problem from Neetcode 150
  • Display AI-generated questions with TTS audio playback
  • End-of-session feedback with strengths, weaknesses, and actionable improvements

• FastAPI Backend

  • POST /chat - Send messages to AI interviewer
  • POST /feedback - Get end-of-session feedback
  • GET /health - Check server and Ollama status
  • Async endpoints for performance

• AI Integration

  • Local Ollama LLM Llama 3.2
  • LeetCode-style interview conversation flow
  • Tracks candidate progress for feedback

• Text To Speech

  • Local Kokoro-onnx TTS for audio generation
  • AI responses converted to speech

Current Development

• Live Kit Integration
• Microphone Input and STT
• General Tutuoring

Setup Instructions

Prerequisites

1. Python 3.10+

   python --version

2. Ollama (Local LLM runtime)

   • Download and install: https://ollama.ai/download

3. Kokoro-onnx TTS

   • Download a Voice Model (~300 MB):

     # adds .onnx model file to backend
     curl -L \
       https://github.com/thewh1teagle/kokoro-onnx/releases/download/model-files-v1.0/kokoro-v1.0.onnx \
       -o backend/voices/kokoro-v1.0.onnx
     

4. Google Chrome (for extension)

---

Backend Setup

Step 1: Install Ollama and Pull Model

# Start Ollama service
ollama serve

# In a new terminal, pull the AI model (only once)
ollama pull llama3.2:latest

# Check downloaded models
ollama list

Step 2: Install Python Dependencies

cd backend

# Create virtual environment (recommended)
python -m venv venv
.\venv\Scripts\Activate.ps1

# Install dependencies
pip install -r requirements.txt

Step 3: Start FastAPI Server

# Make sure you're in the backend directory
cd backend

# Run the server
python app.py

Server will start at: http://localhost:8000

Check health: http://localhost:8000/health

---

Chrome Extension Setup

Step 1: Load Extension in Chrome

1. Open Chrome and go to: chrome://extensions/

2. Enable Developer mode (toggle in top-right)

3. Click "Load unpacked"

4. Navigate to and select the chrome-extension folder

5. The extension should now appear in your extensions list

---

Usage

Starting an Interview

1. Make sure backend is running:

   • Ollama service: ollama serve
   • FastAPI server: python app.py (in backend folder)

2. Open the extension by clicking the icon in Chrome

   • Open the side panel to keep extension open

3. Configure interview:

   • Interview Type: SWE Intern
   • Question Type: LeetCode-style DSA

4. Click "Start Interview"

5. The AI interviewer will:

   • Introduce itself
   • Open Leetcode
   • Describe the first LeetCode-style question
   • Play audio

During the Interview

• Type your approach or solution in the text box
• Ask for hints if you're stuck
• Clarify the question (e.g., "What are the input constraints?")
• The AI will keep track of all questions, and inputs (leetcode code parsing coming soon)

Ending the Interview

1. Click "End Interview" button

2. The AI will generate feedback with:

   • Strengths
   • Weaknesses
   • Actionable Improvement

3. Click "Return to Home Page" to go back to interview setup

---

API Documentation

POST /chat

Send a message to the AI interviewer.

Request:

{
  "messages": [
    {"role": "system", "content": "System prompt"},
    {"role": "user", "content": "User message"}
  ],
  "session_id": "unique-session-id"
}

Response:

{
  "message": "AI response text",
  "audio_url": "/audio/speech_123.wav"
}

POST /feedback

Get end-of-session feedback.

Request:

{
  "messages": [...],
  "session_id": "unique-session-id"
}

Response:

{
  "feedback": {
    "strengths": "Good problem-solving approach...",
    "weaknesses": "Could improve time complexity analysis...",
    "improvement": "Practice more dynamic programming problems"
  }
}

GET /health

Check server health and Ollama connection.

Response:

{
  "status": "healthy",
  "ollama": "connected",
  "sessions": 3
}

---

Dependencies

Backend

• FastAPI - Web framework
• Uvicorn - ASGI server
• httpx - Async HTTP client for Ollama
• pydantic - Data validation
• kokoro-onnx - Text-to-speech

Chrome Extension

• No external dependencies

External Services

• Ollama - Local LLM runtime
• Kokoro-onnx - Local TTS

---

Privacy & Security

Fully Local - All data stays on your machine

• No cloud API calls
• No data collection
• No internet required

Free & Open Source

• All components are free
• No API keys needed
• No subscription costs

---

Happy interviewing!