PrepForge AI — Full-Stack AI-Powered Interview Preparation & Voice Mock Platform

PrepForge AI is a full-stack, end-to-end interview preparation platform that transforms resume parsing and job description analysis into an interactive, voice-driven mock interview experience.

By combining natural language processing, real-time Speech-to-Text (STT), Text-to-Speech (TTS) audio playback, adaptive AI question generation, and empirical performance analytics, PrepForge AI bridges the gap between static studying and real-world interview execution.

---

📋 Table of Contents

• Overview & Architecture
• Key Features
• Tech Stack
• Project Directory Structure
• API Endpoints Reference
• Installation & Setup Guide
• Environment Variables
• Voice Mock Interview Flow
• Progress Analytics & Evaluation
• License

---

🌐 Overview & Architecture

PrepForge AI operates on a modern 4-tier layer architecture:

                          +-----------------------------------+
                          |     Public Landing Page & UI      |
                          |  (React 19, Vite, SCSS, Router 7) |
                          +-----------------------------------+
                                            |
                                            v
                          +-----------------------------------+
                          |  Hooks, Context & Web Speech API  |
                          |  (STT SpeechRec / TTS Synthesis)  |
                          +-----------------------------------+
                                            |
                                            v
                          +-----------------------------------+
                          |  REST API & Middleware Security   |
                          | (Node.js, Express, JWT Cookies)   |
                          +-----------------------------------+
                                            |
                                            v
                          +-----------------------------------+
                          |    AI Engine & Database Layer     |
                          | (Groq GPT-OSS-120B, MongoDB, PDF) |
                          +-----------------------------------+

---

✨ Key Features

🏠 1. Public Landing Page & Product Showcase

• Unauthenticated Entry Point: Introduces visitors to the product workflow, feature capabilities, interactive voice showcase, and progress metrics before requiring authentication.
• Dynamic Header Navigation: Sticky navigation bar offering context-aware actions (Login, Get Started for visitors; Dashboard, Voice Mock, History, Progress, Logout for authenticated users).

🔐 2. Authentication & User Security

• JWT HTTP-Only Cookies: Secure session persistence with strict sameSite: "lax" cookie flags preventing client-side script tampering.
• Token Blacklisting: Server-side token invalidation model upon logout preventing replay attacks.
• User Ownership Isolation: All stored interview plans, voice sessions, and progress analytics are strictly isolated and authorized by user ID.

📄 3. Resume Parsing & Profile Alignment

• PDF Resume Upload: Memory-buffer binary parsing extracting raw resume text using pdf-parse.
• Match Score Engine: Evaluates alignment (0–100%) between the candidate's resume, self-description, and target job description.

🎯 4. Skill Gap Detection & Dynamic Roadmap

• Categorized Skill Gaps: Identifies high, medium, and low severity deficiencies required by the target job role.
• Personalized Preparation Plan: Generates dynamic day-by-day learning roadmaps customized specifically to missing competencies.

💻 5. Technical & Behavioral Q&A Generation

• Role-Specific Q&A: Generates technical and STAR-method behavioral interview questions.
• Interviewer Intention & Answer Guides: Breaks down what the interviewer assesses alongside model answer guidelines.

🎙️ 6. Interactive Voice Mock Interview Engine

• Text-to-Speech (TTS): AI interviewer reads questions aloud using Web Speech Synthesis.
• Real-Time Speech-to-Text (STT): Transcribes candidate spoken answers in real-time using SpeechRecognition / webkitSpeechRecognition.
• Manual Text Fallback: Seamless fallback input for browsers without speech recognition support.
• Adaptive Questioning: AI adjusts question difficulty dynamically:
  • Low Score (<6/10) → Asks simpler/foundational follow-up questions to assess basic knowledge.
  • High Score (>=8/10) → Increases difficulty with complex scenario-based and architectural questions.

📊 7. Instant Evaluation & Feedback

• 0–10 Question Scoring: Scores each answer objectively.
• Strengths & Weaknesses: Extracts key candidate strengths and observed weaknesses.
• Model Answer Generation: Provides ideal responses for comparison.

📜 8. Interview History & Question Transcripts

• Complete Session Storage: Persists all voice interviews in MongoDB.
• Question-Level Breakdown: Detailed review of spoken answer transcripts, scores, AI feedback, and TTS audio replay.

📈 9. Empirical Progress Analytics

• Overall Score Progression: Visual bar chart tracking overall interview performance over time (e.g. Session 1: 58 → Session 2: 67 → Session 3: 74).
• Topic Performance Curves: Tracks improvement across specific domains (SQL, Python, React, System Design, Communication).
• Gap vs. Weakness Comparison: Compares initial initial skill gaps (from resume) against observed interview weaknesses (from spoken performance).

📄 10. ATS-Friendly Resume Generator

• HTML/CSS Resume Compilation: AI tailors resume content specifically to target job keywords without fabrication.
• Puppeteer PDF Export: Renders single-column, standard-compliant ATS PDFs ready for job applications.

---

🛠️ Tech Stack

Frontend

• Framework: React 19, Vite 8
• Routing: React Router 7 (createBrowserRouter)
• Styling: SCSS (Sass 1.102), LightningCSS
• HTTP Client: Axios (with credentials)
• Speech Integration: Web Speech API (SpeechRecognition, SpeechSynthesis)

Backend

• Runtime: Node.js
• Framework: Express 5
• Database: MongoDB & Mongoose 9
• AI SDK: Groq SDK (openai/gpt-oss-120b)
• PDF Processing: pdf-parse & puppeteer
• Security: jsonwebtoken, bcryptjs, cookie-parser, cors

---

📁 Project Directory Structure

PrepForge_AI/
├── Backend/
│   ├── src/
│   │   ├── config/
│   │   │   └── database.js             # MongoDB connection setup
│   │   ├── controllers/
│   │   │   ├── auth.controller.js      # User registration, login, logout, getMe
│   │   │   └── interview.controller.js # Report generation, voice sessions, analytics
│   │   ├── middlewares/
│   │   │   ├── auth.middleware.js      # JWT cookie verification middleware
│   │   │   └── file.middleware.js      # Multer memory storage file upload
│   │   ├── models/
│   │   │   ├── user.model.js           # User schema
│   │   │   ├── blacklist.model.js      # Revoked token schema
│   │   │   ├── interviewReport.model.js# Analysis report schema
│   │   │   └── interviewSession.model.js# Voice mock session schema
│   │   ├── routes/
│   │   │   ├── auth.routes.js          # Authentication routes
│   │   │   └── interview.route.js      # Interview & Voice API routes
│   │   ├── services/
│   │   │   └── ai.service.js           # Groq AI prompts, structured JSON & Puppeteer PDF
│   │   └── app.js                      # Express application initialization
│   ├── .env                            # Backend environment secrets
│   ├── server.js                       # Server entry point (Port 3000)
│   └── package.json
│
├── Frontend/
│   ├── src/
│   │   ├── assets/                     # SVG icons & logo graphics
│   │   ├── features/
│   │   │   ├── auth/
│   │   │   │   ├── components/Protected.jsx
│   │   │   │   ├── hooks/useAuth.js
│   │   │   │   ├── pages/Login.jsx
│   │   │   │   ├── pages/Register.jsx
│   │   │   │   └── auth.context.jsx
│   │   │   ├── interview/
│   │   │   │   ├── components/Navbar.jsx
│   │   │   │   ├── hooks/useInterview.js
│   │   │   │   ├── pages/Home.jsx           # Authenticated Plan Generator & Recent Plans
│   │   │   │   ├── pages/Interview.jsx      # Interview Plan Detail View
│   │   │   │   ├── pages/InterviewVoice.jsx # Interactive Voice Mock Session
│   │   │   │   ├── pages/History.jsx        # Session History & Transcripts
│   │   │   │   ├── pages/Progress.jsx       # Progress Analytics Dashboard
│   │   │   │   ├── pages/LandingPage.jsx    # Public Product Landing Page
│   │   │   │   ├── services/interview.api.js
│   │   │   │   └── interview.context.jsx
│   │   │   └── services/
│   │   │       └── auth.api.js
│   │   ├── App.jsx
│   │   ├── app.routes.jsx              # React Router configuration
│   │   └── main.jsx
│   ├── index.html
│   ├── vite.config.js
│   └── package.json
│
├── .gitignore                          # Git ignore rules
└── README.md                           # Documentation

---

📡 API Endpoints Reference

Authentication Routes (/api/auth)

Method	Endpoint	Access	Description
POST	/api/auth/register	Public	Registers a new user account & sets HTTP-only JWT cookie
POST	/api/auth/login	Public	Authenticates user & sets HTTP-only JWT cookie
POST / GET	/api/auth/logout	Public	Clears cookie & adds token to server blacklist
GET	/api/auth/get-me	Private	Fetches current logged-in user profile

Interview & Voice Routes (/api/interview)

Method	Endpoint	Access	Description
POST	/api/interview	Private	Analyzes resume PDF, self description & JD to create report
GET	/api/interview	Private	Fetches all recent interview plans of logged-in user
GET	/api/interview/report/:interviewId	Private	Fetches detailed interview analysis report by ID
POST	/api/interview/resume/pdf/:interviewReportId	Private	Generates tailored ATS resume PDF via Puppeteer
POST	/api/interview/session/start	Private	Starts a new voice mock interview session
POST	/api/interview/session/:sessionId/answer	Private	Submits spoken transcript, evaluates answer & gets next adaptive Q
POST	/api/interview/session/:sessionId/complete	Private	Finalizes voice mock interview session
GET	/api/interview/sessions	Private	Fetches all past completed voice interview sessions
GET	/api/interview/session/:sessionId	Private	Fetches single voice session details & transcript
GET	/api/interview/progress	Private	Computes overall score progression & topic analytics

---

⚙️ Environment Variables

Create a .env file inside the Backend/ directory:

PORT=3000
MONGO_URI=mongodb+srv://<username>:<password>@<cluster>.mongodb.net/<dbname>
JWT_SECRET=your_jwt_secret_key_here
GROQ_API_KEY=gsk_your_groq_api_key_here

---

🚀 Installation & Setup Guide

Prerequisites

• Node.js (v18+ recommended)
• npm or yarn
• MongoDB Database (Local or MongoDB Atlas)
• Groq API Key

1. Clone Repository

git clone https://github.com/Rohan-shiva/prepforge-ai.git
cd prepforge-ai

2. Backend Setup

cd Backend
npm install

Start the backend server:

npm run dev
# Server running at http://localhost:3000

3. Frontend Setup

Open a new terminal tab:

cd Frontend
npm install

Start the Vite development server:

npm run dev
# Application running at http://localhost:5173

---

🎙️ Voice Mock Interview Flow

       Candidate Clicks "Start Voice Session"
                         │
                         ▼
        AI Interviewer Speaks Question (TTS)
                         │
                         ▼
       Candidate Listens & Speaks Answer (STT)
                         │
                         ▼
       Transcript Displayed & Answer Submitted
                         │
                         ▼
          AI Evaluates Response (0-10)
  (Feedback, Strengths, Weaknesses, Model Answer)
                         │
                         ▼
        Score Check & Dynamic Question Adaptation
       ┌─────────────────┴─────────────────┐
       ▼                                   ▼
Score < 6/10                         Score >= 8/10
(Simpler / Follow-up Q)              (In-depth / Advanced Q)
       └─────────────────┬─────────────────┘
                         │
                         ▼
            Complete 4-Question Session
                         │
                         ▼
       Overall Session Score & Weaknesses Saved

---

📄 License

Distributed under the ISC License. See LICENSE for more information.

---

PrepForge AI — Prepare Smarter. Interview Better. Production Deployment: Active & Verified (Vercel & Render)