Project Overview

PrepAI is a premium, full-stack, intelligence-driven preparation suite designed to help software engineer candidates excel in highly competitive technical recruitment processes. By integrating professional AI tools directly into a cohesive student dashboard, PrepAI provides real-time evaluations across the three core pillars of technical recruiting: resume filtering (ATS), algorithmic coding lab assessments, and live verbal mock technical interviews.

Screenshots

Key Features

1. Interactive Student Prep Dashboard
   • Aggregate Performance Gauge: A visual ring showing the candidate’s dynamic readiness score.
   • 7-Day Performance Trend Chart: A sleek, custom-designed line graph mapping progress over the week.
   • Dynamic AI Recommendation Banner: A glowing advice card pointing out the candidate’s optimal next study step.
   • Chronological Recent Activity Log: An automated chronological list displaying the user’s recent scores, interview completions, and coding submissions.
2. Algorithmic Coding Lab
   • Monaco Code Console: An embedded editor matching professional development environments, supporting JavaScript, Python, C++, and Java.
   • Live Output Stream Panel: A terminal emulator printing compiler diagnostics, test case verification results, and execution statuses.
3. AI Technical Mock Interviewer
   • ** Sarah (AI Interviewer)**: A simulated recruiter who acts as a Google Engineer conducting professional technical and architectural mock interviews.
   • Real-Time Transcript Dialogue: A chat log tracking questions, candidate answers, and contextual feedback.
   • Hardware Controller Toggles: Graphical controls to toggle camera and microphone feeds.
4. ATS Resume Scorer
   • Drag-and-Drop File Uploader: A clean, dashed drop-zone supporting PDF and DOCX uploads under 2MB.
   • ATS Parameter Metrics: Dynamic score indicators evaluating Structure, Keywords, and Impact.
   • Gap Analysis & Skill Recommendations: Targeted lists of critical missing skills and actionable suggestions.
5. Secure Session Authentication Gate
   • Protected Routing Barrier: Gates the Dashboard, Interview, Resume, and Coding Lab behind an authentication wall.
   • Hashed Database Storage: Fully secures passwords inside MongoDB.

Functionaliy

1. Sandboxed JS Code Evaluation
   • Node.js VM Invocation: When a user clicks “Run Code”, their JavaScript code is sent to the backend /api/coding/run endpoint. The server executes it inside a secure virtual machine sandbox (vm module).
   • Timeouts & Safety: Enforces a strict 1000ms execution timeout to neutralize infinite loops.
   • Automated Test Assertions: Evaluates the code against target parameters for the Two Sum problem, checking returned structures, handling runtime errors, and logging results in MongoDB.
2. Document Processing & ATS Scoring
   • Memory Buffer Ingestion: File uploads are captured via multer memory storage.
   • PDF Text Extraction: The raw PDF buffer is parsed using pdf-parse, translating layouts into text strings.
   • Gemini ATS Analysis: The text is sent to Gemini with a structured prompt. Gemini returns a strict JSON object mapping structural, keyword, and impact feedback.
   • Keyword Fallback Matching: If the Gemini API key is missing, a local parser scans the text for key software engineering terms and calculates matching ratios to construct realistic feedback.
3. Conversational AI Chat Flow
   • Conversational Memory: The mock interview page sends the full transcript array to the backend /api/interview/chat route.
   • Gemini Prompt Injection: Injects instructions directing Gemini to act as “Sarah,” an interviewer at Google, maintaining professional interview questions while keeping responses under 3 sentences.
   • Rule-Based Fallback Dialogues: If the Gemini API is offline, a state-aware conversation loop steps through logical questions about React rendering, system design, and security.
   • DB Session Logs: Every conversation is stored in the Interview MongoDB schema under the active user’s ID.
4. Dynamic Dashboard Aggregation
   • Mongoose Pipelines: When a user opens the dashboard, the backend triggers database queries across the Coding, Interview, and Resume collections.
   • Dynamic Score Math:
     • Readiness Score: Calculated dynamically based on resume scores, coding completions, and mock interview attempts.
     • Overall Rank: Shifts dynamically based on the readiness score and solved counts.
     • Prep Time: Calculated based on the candidate’s actual interaction history (e.g., 30 minutes per interview, 15 minutes per coding submission).
5. Secure Cookies Session Management
   • HttpOnly JWT Delivery: When logging in or registering, a signature JSON Web Token is signed with a 30-day expiration and attached to a secure HttpOnly cookie.
   • CSRF & XSS Mitigation: The HttpOnly cookie parameter prevents client-side scripts from reading the token, keeping user sessions secure.
   • Authentication Middleware: The protect middleware interceptor extracts and verifies the cookie token on every restricted API call before authorizing database access.

Technology Used

• Frontend: React 19 & Vite, Vanilla CSS & Tailwind v4
• Backend: Node.js & Express, MongoDB & Mongoose
• Authentication: JWT
• AI Integration: Google Gemini 1.5 Flash API
• File Processing: multer & pdf-parser

Why This Project Is Useful

• Beats the ATS Filter: It analyzes your resume against industry-standard keywords and structures, ensuring your application successfully passes automated applicant filters (ATS) to reach human recruiters.
• Bridges Technical Skills and Communication: Unlike typical coding sites, it combines dynamic coding challenges with professional AI mock interviews, preparing you to both write clean algorithms and verbally defend your design choices.
• Provides Secure, Instant Code Feedback: Runs your code safely on the server inside a secure virtual machine sandbox, testing your logic against real assertion cases with helpful compiler output.
• Offers Realistic AI Recruiting Practice: Simulates a live Google recruiter (“Sarah”) utilizing Google Gemini, challenging your software architecture knowledge in a low-pressure environment.
• Provides a Data-Driven Study Plan: Aggregates all your performance stats and dynamic trends onto a single dashboard, showing you exactly what topics to study next based on your actual performance data.

Deployment and Environment Setup

1. Environment Variables Setup:
   • backend/.env
     • PORT: 3000
     • MONGO_URI: mongodb_url_string
     • JWT_SECRET: your_jwt_secret_key_here (if needed)
   • frontend/.env
     • VITE_API_URL: backend_url
2. Deployment Overview:
   • Render or Railway for backend hosting
   • Vercel or Netlify for frontend hosting
   • MongoDB Atlas for the database
3. Backend Deployment:
   • Push your code to GitHub.
   • Go to Render.com or Railway.app.
   • Create a new Web Service and connect your GitHub repository.
   • Choose “backend/” as the root directory.
   • Add the following environment variables on Render:
   • PORT=10000
   • MONGO_URI=your MongoDB Atlas URI
   • JWT_SECRET=your JWT secret
   • Start command: npm start (Ensure your backend/package.json includes “start”: “node server.js”)
   • After deployment, note your backend URL, e.g. https://your-backend-name.onrender.com
4. Frontend Deployment:
   • Go to Vercel.com or Netlify.com.
   • Create a new project and connect the GitHub repo.
   • Select the “frontend/” folder as root.
   • In environment variables, set:
   • VITE_API_URL=https://your-backend-domain.com/api (or REACT_APP_API_URL if using CRA)
   • Deploy.
5. MongoDB Atlas Setup:
   • Go to https://cloud.mongodb.com
   • Create a free cluster.
   • Add your current IP address in Network Access.
   • Create a database user and password.
   • Copy the connection string and replace username, password, and database name in MONGO_URI in backend/.env

Download Source Code

Project Setup Instructions

Note: This project is for educational purposes only. Not for commercial sale.