the-ats

byPrem Patel

Build a production-ready full-stack AI Resume Analyzer web application. OBJECTIVE Create a complete application where users can create an account, log in, upload a PDF resume, paste a job description, and receive an AI-powered analysis comparing their resume against the job description. The application should be implemented as a real working product, not a visual prototype. Do not use hardcoded fake analysis results. CORE USER FLOW 1. User creates an account or logs in. 2. User uploads a PDF resume. 3. The application securely extracts text from the PDF. 4. User pastes a job description. 5. User starts the analysis. 6. The backend sends the relevant resume and job-description content to an LLM. 7. The application displays a structured analysis. 8. The analysis is saved to the user's account. 9. User can view previous analyses from a dashboard. 10. User can open or delete previous analyses. AI ANALYSIS The analysis must include: - Overall resume-to-job match score from 0–100 - Matching skills - Missing skills - Relevant experience - Potential weaknesses - Recommended resume improvements - Recommended skills to learn - Strengths of the candidate - Concise explanation of why the candidate matches or does not match the job - Clear evidence from the resume supporting important conclusions Do not generate fake/static analysis data. The LLM integration must use an environment variable for the API key. Never expose API keys in frontend code. TECHNICAL REQUIREMENTS Frontend: - React - TypeScript - Tailwind CSS - Responsive design - Reusable components - Clean component architecture Backend: - Backend API - Proper separation between frontend and backend responsibilities - Input validation - Error handling - Secure API key handling - Proper HTTP status codes Database: - PostgreSQL - Store users - Store uploaded resume metadata - Store extracted resume information where appropriate - Store job descriptions - Store analysis results - Maintain proper relationships between users and their analyses - Use appropriate indexes and constraints Authentication: - User registration - Login - Logout - Secure authentication - Protected dashboard and analysis routes - Users must only be able to access their own data PDF PROCESSING: - Accept PDF resume uploads - Validate file type - Validate reasonable file size - Extract actual text from the PDF - Handle invalid or unreadable PDFs gracefully - Do not simply pretend that extraction occurred AI INTEGRATION: - Use a real LLM integration - API key must come from environment variables - Do not hardcode API keys - Structure the AI response in a predictable format - Validate the AI response before displaying/storing it - Handle API failures, malformed responses, rate limits, and timeouts gracefully DASHBOARD Create a professional dashboard where users can: - See their previous resume analyses - See resume/job title information - See match scores - Open a previous analysis - Delete a previous analysis - Start a new analysis UI/UX Create a modern, professional and polished interface suitable for a real SaaS product. Include: - Landing/login experience - Authentication pages - Dashboard - Resume upload interface - Job description input - Analysis loading state - Analysis results page - Previous analysis history - Empty states - Error states - Success states - Confirmation states where appropriate - Mobile responsive layouts - Accessible forms and controls - Clear visual hierarchy The UI should not just look good. All important interactions must actually work. ANALYSIS RESULTS UI Present the analysis in a clear dashboard-style format. Include: - Match score visualization - Skills matched - Skills missing - Experience analysis - Strengths - Weaknesses - Resume improvement recommendations - Learning recommendations - Explanation/evidence Avoid overwhelming the user with one large block of text. SECURITY Implement reasonable production security practices: - Never expose API keys to the client - Validate uploaded files - Validate user input - Protect authenticated routes - Prevent users from accessing another user's analyses - Avoid trusting client-provided user IDs - Handle database errors safely - Do not expose sensitive server information in error messages - Use environment variables for secrets - Apply appropriate authorization checks to every protected resource TESTING Create meaningful tests for important functionality. At minimum test: - Authentication - Protected routes - PDF upload validation - PDF text extraction - Database operations - Analysis creation - Authorization / user isolation - AI response handling - Important API endpoints - Important frontend functionality Run the tests after implementation. If tests fail, investigate and fix the underlying issues rather than simply removing or disabling the tests. PROJECT QUALITY Use: - Clean architecture - Reusable components - Meaningful variable and function names - Type safety - Proper error handling - Separation of concerns - Maintainable code - Minimal unnecessary dependencies Do not generate unnecessary files or complexity. README Create a comprehensive README containing: - Project overview - Features - Architecture - Technology stack - Prerequisites - Installation - Environment variables - Database setup - How to run frontend - How to run backend - How to run tests - How to use the application - Deployment instructions - Troubleshooting Create a `.env.example` file showing all required environment variables without exposing real secrets. IMPLEMENTATION PROCESS Before implementation: 1. Analyze the requirements. 2. Create a clear technical architecture. 3. Identify the required frontend, backend, database and AI components. 4. Create an implementation plan. 5. Identify potential technical risks. Then implement the complete application. After implementation: 1. Install dependencies. 2. Run the application. 3. Run tests. 4. Test the important user flows. 5. Fix errors discovered during testing. 6. Verify authentication. 7. Verify PDF upload and text extraction. 8. Verify database persistence. 9. Verify AI analysis. 10. Verify previous-analysis retrieval. 11. Verify deletion. 12. Verify authorization/user isolation. 13. Verify responsive UI. 14. Verify that the application can be started using the documented instructions. Do not stop after generating code. Continue until the application has been tested and verified as far as the available environment allows. IMPORTANT: If an external API, credential, deployment service, or other resource is unavailable, do not fake the result. Clearly document what could not be verified. DO NOT USE FAKE IMPLEMENTATIONS Do not: - Use hardcoded analysis results - Create fake API responses and claim they are real - Create fake database persistence - Pretend tests passed when they were not run - Pretend deployment succeeded when it was not performed - Claim a feature is complete when it is only a placeholder - Hide errors instead of fixing them If something cannot be implemented because of an unavailable external dependency, clearly mark it as incomplete and explain what is required to finish it. FINAL REPORT REQUIREMENT After completing the implementation and testing, create a file named: FINAL_REPORT.md Place it in the project root. This file is extremely important because it will be used to independently compare this implementation against the same application generated by other AI development platforms. The FINAL_REPORT.md must contain: 1. Project Overview 2. Executive Summary 3. Features Implemented 4. Features Partially Implemented 5. Features Not Implemented 6. Technology Stack 7. System Architecture 8. Complete Project Structure 9. Database Schema 10. API Endpoints 11. Authentication Architecture 12. AI/LLM Integration 13. PDF Upload and Text Extraction Implementation 14. Security Measures 15. Testing Strategy 16. Tests Actually Run 17. Test Results 18. End-to-End Verification Results 19. Errors Encountered 20. Errors Fixed 21. Known Limitations 22. External Dependencies 23. Environment Variables 24. Setup Instructions 25. Run Instructions 26. Deployment Instructions 27. Performance Considerations 28. Future Improvements 29. Overall Completion Assessment For every major feature, clearly classify its status as one of: - IMPLEMENTED - PARTIALLY IMPLEMENTED - MOCKED - NOT IMPLEMENTED - NOT VERIFIED Accuracy is critical. Do not claim that a feature works unless you actually implemented and verified it. For testing, report only tests that were actually executed. For deployment, report only deployment that was actually performed and verified. If something was impossible to verify because credentials, external services, or infrastructure were unavailable, explicitly state: "NOT VERIFIED — [reason]" Do not hide limitations. The FINAL_REPORT.md should provide enough technical detail for another developer to understand exactly what was built, what was tested, what worked, what failed, and what remains incomplete. FINAL DELIVERABLES Before finishing, ensure the project contains: - Complete source code - Working frontend - Working backend - Database implementation - AI/LLM integration - Authentication - PDF processing - Tests - README.md - .env.example - FINAL_REPORT.md Finally, provide a concise summary of: - What was built - What was successfully verified - What remains incomplete - Where the FINAL_REPORT.md file is located Prioritize correctness, functionality, maintainability, security, and honest verification over speed or visual polish.