How rishi-gupta-data/smartscreen works
Project Type and Purpose
• This is a full-stack web application: “SmartScreen ATS”
• Purpose: AI-powered Applicant Tracking System.
• React.js single-page frontend (frontend/)
• Flask REST API backend (backend/)
• ML/NLP engines for resume analysis
• Database for persistent storage
• External AI services (Gemini AI, Google Generative AI API)
File Structure
• Top level
• .devcontainer: VSCode dev-container config
• backend/: Flask app sources (app.py, backend.py, frontend.py) + environment (.env) + docs
• frontend/: Create-React-App structure (public/, src/, package.json)
• requirements.txt: Python dependencies (Flask, Pandas, scikit-learn, PDFMiner, AI SDKs)
• package.json / package-lock.json: Node/React dependencies (React, Tailwind CSS)
• Patterns and clues
• src/components in frontend: UI splits (HomePage, BulkAnalysis, IndividualAnalysis, ResumeAnalysis)
• services/api.js: HTTP client layer
• backend uses ATS_Project_Guide.md and details.md for domain logic
README for Architectural Insights
• User → React Frontend → Flask Backend API → Database + NLP/ML Models
• API endpoints: /api/analyze, /api/bulk-analyze, /api/formatting-suggestions, /api/keyword-optimization
• External integrations: PDFMiner for PDF parsing, Google/Gemini AI for advanced NLP
Final Recommendations
• Be as granular as practical: break the backend into controllers, services, and libraries.
• External dependencies (PDFMiner, AI APIs) separately so they’re not lost in the monolith.
• Database type (SQL/NoSQL)
• Authentication/authorization if applicable
• DevContainer or Docker build pipeline if infrastructure context is needed