How krishnanshu-khanna/chat-resume-ai works
Project type and purpose
• CVHelper is a full-stack resume builder web application built with Next.js.
• It provides AI-powered features: resume editing, ATS-optimized chat, interview practice, PDF parsing/generation, user auth, and paid subscriptions.
High-level components
Module-wise (to fit a portrait-oriented canvas), group the system into these main blocks:
a. Client (Browser) – Next.js React components
b. Next.js Application Layer – server-side rendering & API routes
c. Data Layer – Prisma + PostgreSQL
d. External Services – Clerk (Auth), Stripe (Payments), AI APIs (OpenAI, Gemini), Langchain, Pinecone, Vercel Neon Blob
Relationships & data flows
• The browser loads pages and invokes React UI components in src/app and src/components.
• UI triggers Next.js server (via page requests, app router, and app/api routes).
• The Next.js server communicates with:
• Prisma client → Postgres database (user data, resumes, chats, interviews, subscriptions)
• Clerk SDK (authentication middleware)
• Stripe API & stripe-webhook route (billing management)
• File upload route (/api/upload) → Blob storage (Vercel Neon Blob)
• AI modules in src/lib (langchain + pinecone for ATS chat, Gemini/OpenAI for text generation)
Architectural patterns & design principles
• Componentized UI (atomic + module-level components under src/components and src/app/(main)).
• Modular services per feature: auth, billing, chat, interview, editor.
• Serverless functions / API routes for integration points (Stripe webhooks, file uploads).
• Prisma ORM for data modeling and type safety.
• Third-party SDKs isolated in src/lib.