Placement-Pro

Placement-Pro is an AI-assisted career application platform built to help job seekers improve resume quality, measure role fit, generate recruiter outreach, and track application activity in one workflow.

This project started as a set of resume tools and was gradually shaped into a more product-focused experience with separate ATS checking, JD matching, cold email generation, bullet improvement, and application history.

Highlights

• Built a full-stack AI product around a real job-search workflow instead of isolated generators
• Separated ATS checking from JD matching to reflect two different user problems
• Moved AI prompt logic from the frontend into protected backend routes
• Added authenticated application history to support repeated usage and future product scaling
• Designed the backend to run locally and in a Vercel-style deployment setup

What Problem It Solves

Many candidates apply with the same resume to every role and do not know:

• whether their resume is ATS-friendly
• how well their resume matches a specific job description
• what keywords or skills are missing
• how to write better project bullets
• how to send a stronger recruiter outreach email

Placement-Pro addresses those gaps with a guided application workflow instead of isolated utilities.

Current Product Scope

Phase 1 features implemented:

• ATS Resume Checker Evaluates general resume quality, structure, and ATS readiness
• JD Matching Tool Compares a resume against a target job description and returns match score, missing keywords, and advice
• Cold Email Generator Creates a personalized recruiter outreach email using resume and job context
• Bullet Improvement Converts rough project descriptions into stronger resume-ready bullet points
• Application History Saves ATS checks, JD matches, email drafts, and bullet improvement activity per user

Key Technical Decisions

• AI prompt logic is handled in the backend, not the frontend
• Gemini API is used through protected Express routes
• JWT-based authentication protects personalized features and saved history
• MongoDB stores user data and application activity history
• The backend is structured to run both locally and on Vercel

Tech Stack

• Frontend: React, React Router, Axios, Tailwind CSS
• Backend: Node.js, Express, MongoDB, Mongoose, JWT, Multer, pdf-parse
• AI: Gemini API
• Deployment: Vercel-ready backend setup with env-based CORS configuration

Main User Flow

1. User signs up or logs in
2. User checks resume with the ATS Resume Checker
3. User compares the same resume with a specific job in the JD Matching Tool
4. User improves project bullets if needed
5. User generates a recruiter cold email
6. All major actions are stored in Application History

Architecture Overview

frontend/
  React client
  protected routes
  feature pages for ATS, JD match, email, bullets, history

backend/
  Express API
  auth routes
  AI routes
  Gemini-based controller logic
  MongoDB models

Core API Endpoints

All AI endpoints require authentication.

• POST /api/resume/ats-check
• POST /api/resume/jd-match
• POST /api/email/generate
• POST /api/bullets/generate
• GET /api/auth/history
• GET /api/auth/profile
• PUT /api/auth/profile

Environment Variables

Create backend/.env with:

PORT=8000
MONGODB_URI=your_mongodb_connection_string
JWT_SECRET=your_jwt_secret
GEMINI_API_KEY=your_gemini_api_key
GEMINI_MODEL=gemini-2.5-flash
CORS_ORIGINS=http://localhost:3000,https://your-frontend-domain.vercel.app

Run Locally

Backend

cd backend
npm install
npm start

For development with auto-reload:

cd backend
npm run dev

Frontend

cd frontend
npm install
npm start

If needed, set REACT_APP_API_URL in the frontend environment.

Why This Project Is Worth Reviewing

• It reflects product thinking, not just API integration
• It separates general ATS checking from role-specific JD matching
• It stores user activity to support repeat usage and future product expansion
• It moves AI business logic to the backend instead of exposing prompts in the client
• It demonstrates full-stack work across authentication, AI integration, stateful workflows, and deployment concerns

Challenges Solved

• Refactored the project from frontend-owned prompt generation to backend-owned AI orchestration
• Split one overloaded analysis tool into two clearer product features: ATS checking and JD matching
• Added shared history tracking across multiple AI features instead of storing only scan scores
• Updated backend startup and CORS handling to better support both local development and deployment
• Reworked the project from a tool collection into a more realistic user journey

What I Would Improve Next

• Create a dedicated Application model instead of relying only on activity history
• Add resume versioning for different job applications
• Add background jobs and queueing for heavier AI and document-processing tasks
• Introduce usage limits, analytics, and plan-based feature controls
• Build recruiter or placement-cell dashboards for a stronger B2B use case

Next Product Steps

If extended further, the strongest next additions would be:

• saved applications as first-class records
• resume versioning per job
• interview question generation from a JD
• recruiter dashboard or placement-cell analytics
• usage limits and subscription plans