PrepLens: Interview Management System

The Dev Challenge 2025: Recognized as a Winner in the hackathon

Welcome to the AI-Powered Interview Preparation Platform! This innovative application bridges the gap between traditional placement preparation and real-world interview success by simulating realistic interviews, offering personalized feedback, evaluating resumes, and providing targeted technical training.

---

Resume review

Face analysis

Feedback generation

---

Features

1. Realistic Interview Simulations

• Question Generation: Generates interview questions based on job descriptions.
• Speech-to-Text Conversion: Records and transcribes user responses using React Speech-to-Text.
• Feedback & Rating: Provides real-time feedback and ratings for each answer using Gemini AI.

2. Resume Analysis

• ATS Scoring: Analyzes uploaded resumes and generates ATS scores.
• Keyword Suggestions: Identifies missing keywords relevant to the job description.

3. Face and Posture Analysis

• Real-Time Feedback: Monitors user’s face and posture during mock interviews using Face-API.js.
• Alert System: Alerts: users if they move out of the webcam frame.

4. Comprehensive Feedback

• Detailed Reports: Summarizes performance with actionable insights.
• Confidence Building: Enhances interview readiness with iterative practice and personalized guidance.

---

Tech Stack

Frontend & Backend

Next.js: Framework for building the user interface.
TypeScript: Ensures type safety and scalability.
ShadCN-UI: Provides modern and accessible UI components.
react-hook-peech-to-text: Powers the speech recognition feature.

AI & APIs

Gemini AI: Used for question generation, feedback, and resume ATS scoring.
Face-API.js: Powers real-time facial and posture analysis.

Other Tools

Prisma: Manages database interactions.
Supabase: Handles authentication and storage.
Framer Motion: Adds animations for a dynamic user experience.

---

Prerequisites

1. Node.js (>= 16.x)
2. Environment variables configured (see .env section below).

---

Installation

1. Clone the Repository

# Clone this repository
git clone https://github.com/abhi9ab/Interview-Preparation-App.git

# Navigate into the project directory
cd Interview-Preparation-App

2. Install Dependencies

npm install

3. Configure Environment Variables

Create a .env file in the root of the project with the following variables:

NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=
CLERK_SECRET_KEY=

NEXT_PUBLIC_CLERK_SIGN_IN_URL=/sign-in
NEXT_PUBLIC_CLERK_SIGN_UP_URL=/sign-up

NEXT_PUBLIC_GEMINI_API_KEY=

NEXT_PUBLIC_INTERVIEW_QUESTION_COUNT=5

# Connect to Supabase via connection pooling with Supavisor.
DATABASE_URL=

# Direct connection to the database. Used for migrations.
DIRECT_URL=
        
NEXT_PUBLIC_GOOGLE_API_KEY=

4. Set Up Prisma

1. Generate the Prisma Client:

npx prisma generate

2. Apply migrations to the database:

npx prisma migrate dev --name init

5. Run the Development Server

npm run dev

The application will now be available at http://localhost:3000.

---

API Keys Needed

This project requires the following API keys:

1. Clerk:

   • NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY: Your Clerk Publishable Key.
   • CLERK_SECRET_KEY: Your Clerk Secret Key.

   To obtain these keys:

   • Sign up at Clerk.dev.
   • Create a new application and retrieve the required keys from the dashboard.

2. Gemini-Flash-1.5-8b API Key:

   • NEXT_PUBLIC_GEMINI_API_KEY: API key for accessing the Gemini model.

3. Google Speech-to-Text API Key:

   • NEXT_PUBLIC_GOOGLE_API_KEY: API key for speech recognition.

---

Contributing

Contributions are welcome! Please follow these steps:

• Fork the repository.
• Create a new branch for your feature or bugfix.
• Commit your changes and push to the branch.
• Submit a pull request.

---

Deployment

1. Use a platform like Vercel for deploying the Next.js app.
2. Ensure environment variables are set up in the hosting provider’s dashboard.
3. Use a cloud database service like AWS RDS or Supabase for PostgreSQL.

---

Contact

For further queries, please contact abhinabdas004@gmail.com.

License

This project is licensed under the MIT License. See the LICENSE file for details.