How navaneetha-krishnan-mv/resume-analyzer works
Project type & purpose (what you’re diagramming)
There is also a Python PDF extractor (python/pdfextractor.py) which likely performs PDF-to-text extraction either:
• as a local/dev tool, or
• as code intended to run in a serverless function/container (Lambda) for parsing resumes before AI scoring.
• a frontend web app (this repo),
• a serverless analysis backend (conceptual/external in this repo),
• storage and AI services (conceptual/external),
• optional client-side storage (IndexedDB) and deployment (Vercel).
Frontend (React SPA)
• Entry points / app shell:
• index.html (Vite root HTML)
• src/main.jsx (bootstraps React)
• src/App.jsx (top-level routing/layout composition)
• Pages (feature-level UI screens):
• src/pages/Home.jsx
• src/pages/Upload.jsx (resume ingestion)
• src/pages/Analyze.jsx (analysis results/progress)
• UI Components:
• src/components/Layout.jsx (shared layout/chrome)
• src/components/ProgressBar.jsx (analysis progress UI)
• src/components/ScoreRing.jsx (score visualization)
• src/components/AiScannerLogo.jsx (branding)
• Client-side persistence:
• src/utils/indexedDb.js (IndexedDB usage—likely caching uploads/results/state)
• Tailwind: tailwind.config.cjs, src/index.css
• Vite: vite.config.js
• Linting: eslint.config.js
Hosting/Deployment
• vercel.json strongly suggests hosting the SPA on Vercel (static + edge routing).
• public/ contains static assets shipped with the frontend bundle.
Python utility / parsing stage
• python/pdfextractor.py indicates a PDF extraction stage that should appear in the backend pipeline (even if implemented elsewhere in prod).
What the README tells you
• React frontend
• AWS serverless backend boundary
• AI analysis service(s)
• Data flow from upload → extraction → AI scoring → results returned
Upload
• Browser → SPA → API (upload request)
• API → Object storage
Extract
• Compute → PDF extractor (Python)
• Extractor output → orchestration
Analyze
• Orchestrator → AI provider
• AI response → results store
Return
• SPA polls or receives callback (if polling, show SPA → API “get status/results”)
• API → SPA returns scoring + feedback
• SPA caches in IndexedDB and renders ScoreRing/ProgressBar
Goal: reflect what’s actually here.
Inside “React SPA”, show:
• Router/App Shell: App.jsx
• Pages:
• Home
• Upload (calls API; writes IndexedDB)
• Analyze (reads IndexedDB; calls API for status/results; renders progress + scores)
• Shared UI:
• Layout, ProgressBar, ScoreRing
• Utility:
• indexedDb.js
• Upload/Analyze pages ↔ IndexedDB utility
• Page → ScoreRing/ProgressBar components (data-driven rendering)
Because you have Vercel + Vite:
• Developer → Git repo → Vercel build (Vite) → CDN/Edge hosting
If AWS IaC isn’t present, keep AWS deployment generic.
Relationships & interactions you must show (data flow)
• Resume file (PDF/DOCX): from user → SPA → backend (upload)
• Extracted text: produced by PDF extractor → AI scoring
• Prompts / model inputs: orchestrator → AI provider
• Scores + suggestions: AI provider → orchestrator → results store → SPA
• Progress updates: SPA ↔ API (polling) OR backend → SPA (webhook/websocket)
Architectural patterns to highlight (specific to this repo)
• SPA + Serverless backend: thin UI, stateless API, scalable compute
• Pipeline-style processing: upload → extract → analyze → persist → present
• Client-side caching: IndexedDB to persist state across refresh/navigation
• Presentation components: ScoreRing/ProgressBar indicate a results visualization layer
• Static hosting: Vercel/CDN serving compiled assets
• Clear labels: “Upload Resume”, “Extract Text (Python)”, “AI Scoring”, “Results DB”, “Object Storage”
• “PDF bytes”
• “extracted plain text”
• “analysis requestId”
• “scores/feedback JSON”
• Boundaries:
• “Browser”
• “Vercel (hosting)”
• “AWS account / VPC (if relevant)”
• “External AI Provider”
• Rectangles = compute/services (SPA, API, Lambdas)
• Cloud boundary = AWS / external providers