🤖 Interview AI

Interview AI is an intelligent, full-stack mock interview and preparation platform designed to help candidates prepare for their target roles. By analyzing a candidate's uploaded resume (PDF), self-description, and target job description, the application leverages the Google Gemini API to generate comprehensive, structured feedback reports, identify critical skill gaps, construct custom technical & behavioral questions, formulate study roadmaps, and build downloadable, ATS-friendly tailored resumes.

---

✨ Key Features

• 🔐 Secure Authentication: JWT-based user registration, login, and protected routing.
• 📄 Resume Parse & Analysis: Automated text extraction from PDF resumes using pdf-parse.
• 📊 Compatibility Match Score: Instant calculation of a percentage match score (0-100%) against the target job description.
• 🧠 Custom Technical & Behavioral Q&As: Tailored mock questions complete with the interviewer's hidden intentions and detailed guide answers.
• 🎯 Skill Gap Detection: Identification of missing skills categorized by severity level (low, medium, high).
• 📅 7-Day Actionable Prep Roadmap: Day-by-day structured learning path customized to fill identified skill gaps and master target topics.
• 📄 Tailored ATS-Friendly Resume PDF Generator: Custom-tailored HTML resume structured by Gemini and exported directly to a downloadable PDF via puppeteer.
• 📂 Historical Reports Dashboard: Track progress and revisit previous mock interview analysis sessions at any time.

---

🛠️ Tech Stack

Frontend

• Framework: React + Vite
• Routing: React Router v6 (Browser Router)
• State Management: React Context API
• Styling: SCSS (Sass) for component-level modular styles and utility classes
• HTTP Client: Axios

Backend

• Runtime: Node.js
• Framework: Express.js
• Database: MongoDB with Mongoose ODM
• Authentication: Cookie-based JWT tokens & Bcryptjs password hashing
• AI Integration: Google Gemini SDK (@google/genai) using gemini-3-flash-preview
• File Parsing: multer (multipart/form-data upload) & pdf-parse (PDF text extraction)
• PDF Engine: Puppeteer (Headless browser rendering for resume generation)
• Schema Validation: Zod & zod-to-json-schema

---

🏗️ Folder Structure

interview-ai/
├── Backend/
│   ├── src/
│   │   ├── config/             # DB & server configurations
│   │   │   └── db.js
│   │   ├── controller/         # Request handling logic (Auth & Interview reports)
│   │   │   ├── auth.controller.js
│   │   │   └── interviewController.js
│   │   ├── middlewares/        # JWT auth, File uploads (Multer)
│   │   │   ├── auth.middleware.js
│   │   │   └── file.middleware.js
│   │   ├── models/             # Mongoose schemas (User & InterviewReport)
│   │   │   ├── interviewReport.model.js
│   │   │   └── user.model.js
│   │   ├── routes/             # Router declarations
│   │   │   ├── auth.route.js
│   │   │   └── intervirew.route.js
│   │   ├── services/           # Gemini AI calls & Puppeteer PDF rendering
│   │   │   └── ai.service.js
│   │   └── app.js              # Express app setup & CORS config
│   ├── .env                    # Backend environment configurations
│   ├── server.js               # Entry point
│   └── package.json
├── Frontend/
│   ├── public/                 # Static assets (Favicons, Banner, SVGs)
│   │   └── banner.png
│   ├── src/
│   │   ├── assets/             # Images & static assets
│   │   ├── features/           # Modularized features
│   │   │   ├── Auth/           # Pages & Components for Login & Registration
│   │   │   └── interview/      # Home, Dashboard, Detail page & API services
│   │   │       ├── hooks/      # useInterview custom hook
│   │   │       ├── pages/      # Home.jsx, Interview.jsx
│   │   │       ├── services/   # Axios API endpoints integration
│   │   │       └── interview.context.jsx
│   │   ├── style/              # Global variables, mixins, and stylesheets
│   │   ├── App.jsx
│   │   ├── app.routes.jsx      # React router routing tree
│   │   └── main.jsx
│   ├── package.json
│   └── vite.config.js
└── README.md

---

🚀 Getting Started

Follow these steps to set up the project locally on your machine.

Prerequisites

• Node.js (v18 or higher recommended)
• MongoDB (Local installation or MongoDB Atlas cluster)
• Google Gemini API Key

---

Step 1: Clone the Repository

git clone https://github.com/your-username/interview-ai.git
cd interview-ai

---

Step 2: Configure the Backend

1. Navigate to the Backend folder:

   cd Backend

2. Install dependencies:

   npm install

3. Create a .env file in the Backend/ directory and configure the environment variables:

   MONGODB_URI=your_mongodb_connection_string
   JWT_SECRET=your_jwt_secret_key
   GOOGLE_GEMINI_API_KEY=your_gemini_api_key

4. Start the backend server in development mode:

   npm run dev

   The server runs by default on http://localhost:3000.

---

Step 3: Configure the Frontend

1. Navigate to the Frontend folder:

   cd ../Frontend

2. Install dependencies:

   npm install

3. Start the Vite React development server:

   npm run dev

   The frontend client runs by default on http://localhost:5173.

---

📡 API Reference

Authentication Endpoints (/api/auth)

Method	Endpoint	Description	Access
POST	/register	Register a new user	Public
POST	/login	Log in user, returns HttpOnly cookie	Public
GET	/logout	Invalidate cookie and logout	Public
GET	/get-me	Get currently logged-in user profile	Private (JWT)

Interview Prep Endpoints (/api)

Method	Endpoint	Description	Access
POST	/interview	Upload resume PDF, self description & target job description to generate AI report	Private (JWT)
GET	/interview/:interviewId	Retrieve details for a specific interview report	Private (JWT)
GET	/interviews	Retrieve list of all reports (metadata only) generated by the current user	Private (JWT)
POST	/interview/resume/pdf/:interviewReportId	Generate and download an ATS-optimized, PDF tailored resume	Private (JWT)

---

🤖 Google Gemini Integration

The AI features use the Google Gemini API (gemini-3-flash-preview) to process multi-format context.

Report Schema Structure

To ensure responses are parsed correctly on the frontend, Gemini is forced to output structured JSON data conforming to the following structure:

• matchScore (number): Compatibility score between 0 and 100.
• title (string): Extracted target role title.
• technicalQuestions (Array): Array of objects with question, intention, and answer.
• behavioralQuestions (Array): Array of objects with question, intention, and answer.
• skillGaps (Array): Missing skill tags alongside severity level (low, medium, high).
• preparationPlan (Array): Day-by-day (1-7) structured goals and specific tasks.

Tailored Resume Generation

When downloading the resume:

1. The user's parsed resume, self-description, and target job description are fed to Gemini.
2. The AI generates a clean, custom HTML structure specifically tailored to match the target job description while maintaining an ATS-friendly format.
3. The backend uses Puppeteer to spin up a headless browser, render the custom HTML content, and print it to a high-quality A4 PDF format.