AI Interviewer Platform

An end-to-end, production-ready AI Interviewer Platform that automates the technical and behavioral interview process. The platform features a real-time AI interviewer capable of speech interaction, vision-based behavior analysis, and dynamic question generation.

🌟 Key Features

• Real-time AI Interviewer: Voice-enabled interaction using local Whisper (STT) and Edge-TTS.
• Dynamic Questioning: Adaptive difficulty scaling based on candidate performance.
• Vision Analysis: Behavioral tracking (eye contact, head movement) using MediaPipe.
• Resume-Driven Interviews: Personalized interview focus based on uploaded resumes.
• Comprehensive Reports: Detailed feedback with performance metrics and video recording.
• Admin Dashboard: Manage interviews, candidates, and review performance reports.
• Fully Local AI Option: Support for local LLM inference via Ollama (Llama 3.1).

🏗️ System Architecture

The platform consists of three main services:

1. Frontend (Client): A Next.js 16 application for students and admins.
2. Backend (Server): An Express 5 API server handling business logic, authentication, and data orchestration.
3. AI Service: A Python FastAPI service managing LLM reasoning, STT, TTS, and vision analysis.

---

🛠️ Tech Stack

Frontend (Client)

• Framework: Next.js 16 (App Router)
• Styling: Tailwind CSS 4
• Components: Radix UI, Lucide React, Framer Motion
• Editor: Monaco Editor (for technical coding rounds)
• State/Data: Axios, Zod, React Hook Form

Backend (Server)

• Language: TypeScript
• Framework: Express 5
• Database: MongoDB (via Mongoose)
• Real-time: Socket.io
• Payments: Stripe Integration
• Mailing: Nodemailer

AI Service

• Framework: FastAPI
• LLM Engine: Ollama (Local Llama 3.1:8b)
• Speech-to-Text: OpenAI Whisper (Local)
• Text-to-Speech: Edge-TTS / gTTS
• Vision: MediaPipe (Head pose & eye tracking)
• Data Processing: NumPy, Librosa, OpenCV

---

🚀 Getting Started

Prerequisites

• Node.js: v18+
• Python: v3.10+
• MongoDB: v6+ (or use Docker)
• Ollama: Installed and running (for local LLM)

1. AI Service Setup

1. Navigate to ai-service:

   cd ai-service

2. Create and activate a virtual environment:

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

3. Install dependencies:

   pip install -r requirements.txt

4. Download the LLM model:

   ollama pull llama3.1:8b

5. Start the service:

   python -m app.main

   The service runs at http://localhost:8000

2. Backend Server Setup

1. Navigate to server:

   cd server

2. Install dependencies:

   npm install

3. Configure environment variables (copy .env.example if available or create .env):
   • PORT=5001
   • MONGO_URI: Your MongoDB connection string
   • JWT_SECRET: A secure random string
4. Start development server:

   npm run dev

3. Frontend Client Setup

1. Navigate to client:

   cd client

2. Install dependencies:

   npm install

3. Start development server:

   npm run dev

   Client will be available at http://localhost:3000

---

🐋 Database (Docker)

To quickly start a MongoDB instance:

docker-compose up -d

📄 License

This project is licensed under the ISC License.