PrepAI

PrepAI is a full-stack AI mock interview platform. It lets a user upload a resume, generate role-focused interview questions, answer by text or voice, get per-question scoring/feedback, and review a final analytics report.

Highlights

• Resume upload and parsing (PDF only)
• AI-generated interview questions (configurable count: 1 to 20)
• Domain-based interview tracks:
  • Frontend Developer
  • Backend Developer
  • Data Structures
  • HR Interview
  • System Design
  • Employee Introduction
• Answer evaluation with score + structured feedback
• Voice answer mode with live waveform/intensity
• Confidence metrics (client-side) from transcript/audio stats
• Final report with score trends and weak-topic analysis
• Auth (register/login/me) with JWT
• Resume persistence for logged-in users

Tech Stack

• Frontend: React 19, React Router 7, Framer Motion, Axios, Vite, Tailwind CSS v4, react-loading-skeleton
• Backend: Node.js, Express 5, Mongoose, Multer, pdf-parse, Axios
• Database: MongoDB
• AI Layer:
  • Chat completion via xAI/Groq-compatible OpenAI-style endpoint
  • STT via Groq-compatible transcription endpoint
  • TTS endpoint currently returns 501 and frontend falls back to browser speech synthesis

Monorepo Structure

interview-prep-ai/
  backend/
    scripts/
      check-routes.js
    src/
      config/
      controllers/
      middleware/
      models/
      routes/
      services/
  frontend/
    public/
    src/
      assets/
      components/
      context/
      pages/
      services/

Application Flow

1. User uploads resume PDF on /resume.
2. Backend extracts text from PDF and calls AI parser.
3. Parsed resume is saved locally in browser; if authenticated, it is also saved in MongoDB (Resume model).
4. User starts interview on /interview.
5. Backend generates questions from parsed resume + selected domain.
6. User submits text or voice answers.
7. Each answer is evaluated and scored (0-10) and stored in Interview.responses.
8. User finishes interview and navigates to /report for final analytics.

Frontend Pages

• / Home page with animated landing sections
• /signup Register/Login page
• /resume Resume upload + parsed preview
• /interview Interview engine (question flow, text/voice answer, waveform, auto-read)
• /report Final report + historical progress analytics

Routes are lazy-loaded with Suspense fallback skeleton UI.

Voice + Speech Behavior

• Voice input:
  • Uses MediaRecorder in browser
  • Sends audio to /api/test/stt
  • If route not found (404), retries /api/interview/voice-answer
• STT support:
  • Available only when provider resolves to Groq-compatible transcription in current backend setup
• Question reading (TTS):
  • Frontend first tries /api/test/tts with a short soft-wait
  • If unavailable/slow, it falls back to browser speechSynthesis
  • TTS responses are cached in-memory per question text on the page

API Reference

Base backend URL (local): http://localhost:5000

Notes:

• All routes are mounted under /api/*
• Backward-compatible aliases also exist without /api (for example /auth/login)

Health

• GET / -> returns backend running message

Auth

• POST /api/auth/signup -> create account
• POST /api/auth/register -> alias for signup
• POST /api/auth/login -> login
• GET /api/auth/me -> current user (requires Bearer token)

Resume

• POST /api/resume/upload -> upload PDF (multipart/form-data, field resume)
  • Optional auth: if token present, parsed resume is saved to DB for user
• GET /api/resume/me -> fetch saved parsed resume (auth required)
• DELETE /api/resume/me -> delete saved parsed resume (auth required)

Interview

• POST /api/interview/start -> start interview session
• POST /api/interview/generate -> alias for start
• POST /api/interview/answer -> submit answer
• POST /api/interview/:interviewId/answer -> submit answer (param variant)
• POST /api/interview/evaluate -> alias for answer
• POST /api/interview/finish -> finalize interview and get final score/report
• POST /api/interview/voice-answer -> voice transcription alias (multipart audio)

Test Utilities

• POST /api/test/stt -> speech-to-text transcription (multipart/form-data, field audio)
• POST /api/test/tts -> currently returns 501 in this integration

Key Request Shapes

Start Interview

POST /api/interview/start

{
  "parsedResume": { "skills": ["..."] },
  "domain": "Frontend Developer",
  "questionCount": 3
}

Accepted resume payload aliases: parsedResume, parsedData, resumeData, resumeText.

Submit Answer

POST /api/interview/answer

{
  "interviewId": "mongo_object_id",
  "answer": "Your answer here"
}

Finish Interview

POST /api/interview/finish

{
  "interviewId": "mongo_object_id"
}

Data Models (MongoDB)

• User
  • name, optional unique username, optional unique email, password (hashed)
• Resume
  • one-per-user (user unique), parsedData
• Interview
  • user (optional), resumeData, domain, questions[], responses[], currentQuestionIndex, finalScore, finalFeedback

Local Storage Keys Used by Frontend

• prepai-auth-token
• prepai-user
• prepai-remember-identity
• parsedResume
• finalResult
• interviewHistory
• selectedInterviewDomain
• selectedInterviewQuestionCount

Setup (Local Development)

Prerequisites

• Node.js 20+ recommended
• npm
• MongoDB instance (local or cloud)

1) Clone

git clone https://github.com/Ankit052003/Prep_AI.git
cd Prep_AI

2) Backend

cd backend
npm install
npm run dev

Backend runs on http://localhost:5000 by default.

3) Frontend

Open a second terminal:

cd frontend
npm install
npm run dev

Frontend runs on http://localhost:5173 by default.

Environment Configuration Notes

This README intentionally omits API-key values/details.

• Backend requires database/auth/AI provider configuration in backend/.env.
• Frontend supports optional env overrides for API base/proxy target/media URLs.
• Vite proxy target defaults to http://localhost:5000 in frontend/vite.config.ts.
• Frontend API client default base is http://localhost:5000/api if no override is provided.

Scripts

Backend (backend/package.json)

• npm run dev -> start with nodemon
• npm start -> start with node
• npm run check:routes -> route existence/status check script

Frontend (frontend/package.json)

• npm run dev -> start Vite dev server
• npm run build -> type-check + production build
• npm run lint -> ESLint
• npm run preview -> preview production build

Current Constraints / Known Behavior

• Backend TTS endpoint (/api/test/tts) currently returns 501; browser speech synthesis is used as fallback.
• STT is wired for Groq-compatible transcription in this backend flow.
• Resume upload accepts PDF only.
• Interview question generation defaults to 1 question if not specified.

Troubleshooting

• Backend fails on startup:
  • Verify MongoDB is reachable and backend env config is present.
• 401 on protected routes:
  • Ensure Bearer token is sent from frontend auth session.
• Voice transcription not working:
  • Confirm browser microphone permission is granted.
  • Check /api/test/stt availability and provider compatibility.
• Slow/failed question audio reading:
  • App automatically falls back to browser TTS if backend TTS is unavailable.

License

MIT License. See LICENSE.