VAPI AI Voice Assistant for Candidate Interviews

This project is a full-stack AI voice assistant application built to streamline and automate the interview process for candidate screening. It uses Flask and PostgreSQL on the backend, React with Vite on the frontend, and integrates the Vapi.ai platform for real-time voice-based conversations.

🧠 What Problem Does It Solve?

Traditional interviews are time-consuming and resource-intensive. This assistant simplifies the screening process by:

• Conducting AI-powered interviews in real-time.
• Automatically summarizing key points from conversations.
• Allowing admins to review and manage candidate responses from a dashboard.

⚙️ Technologies Used

• Frontend: React (Vite), CSS, Vapi Web SDK
• Backend: python, Flask, SQLAlchemy, JWT, PostgreSQL
• AI Integration: Vapi.ai platform

🚀 Features

• User registration and login system (JWT)
• Admin access (predefined)
• Voice interview via AI agent (Vapi)
• Summary display post-interview
• Interview history (WIP)
• Admin dashboard to manage users (WIP)
• Dataset download for all registered users with their interview summary
• Admin decision for each candidate (accept/deny/postpone)

📁 Project Structure

├── backend            # Flask backend API
├── frontend           # React frontend application
├── structure.txt      # Project overview and notes
└── README.md

🛠️ Getting Started

1. Clone the Repository

git clone https://github.com/Houssam-11/VoIP-AI-assistant.git
cd VAPI-AI-Voice-Assistant

2. Backend Setup

cd backend
python -m venv env
source env/bin/activate  # On Windows: env\\Scripts\\activate
pip install -r requirements.txt  # or install manually: flask, flask_sqlalchemy, flask_jwt_extended, psycopg2-binary

add .env file where you store Secret_keys to connect with vapi private api, configure postgres credentials, and add JWT_secret_key :

VAPI_API_KEY="VAPI_PRIVATE_API_KEY"
JWT_KEY="YOUR_JWT_KEY" # you can get one by running : "openssl rand -base64 32" in your terminal 
DATABASE_URI="postgresql://name:password@localhost:5432/db_name"

then start the backend server:

python main.py

3. Frontend Setup

create another .env in your frontend folder :

VITE_VAPI_API_KEY="VAPI_PUBLIC_API_KEY"
VITE_ASSISTANT_ID="YOUR_ASSISTANT_ID"

in another terminal run :

cd ../frontend
npm install
npm run dev

The frontend will be available at: http://localhost:5173

🔑 Vapi Platform Setup

1. Create an account at https://vapi.ai
2. Get your API key from the dashboard
3. Update .env in your frontend project with your Vapi credentials

✅ Usage

• Register as a user and log in with your credentials
• Start a voice interview with the AI assistant
• Fill in the form to start the call
• After the call, view a summary of the conversation
• Admins can access a dashboard (WIP) to view user data

• Admins can make decision to accept/deny or postpone a candidate application
• Admins can download a dataset of all registered candidates and their call summary

🤝 Contributing

Feel free to fork this repo and submit pull requests. Suggestions and improvements are welcome!

🙏🏼 Acknowledgement

This Project was developed in Orange Digital Center Rabat, Maroc. thanks to all members who contributed in this project!