How utsavbhardwaj/interviewai works
Project type and purpose
• This is a full-stack web application (“Interview AI”) for practicing mock interviews using AI-generated questions and real-time voice/video interaction.
• Frontend: React + TypeScript, Vite, Tailwind CSS, Shadcn/UI, Wouter, TanStack Query, browser media APIs.
• Backend: Node.js + Express in TypeScript, Drizzle ORM + PostgreSQL (Neon), Multer for uploads, session store via connect-pg-simple, custom storage, AI services integration (Gemini, Deepseek).
File structure
• Top-level “client” and “server” directories indicate a clear frontend/backend separation.
• “client/src/components” subdivides UI, feedback, interview modules—suggests component-based design.
• “client/src/hooks” encapsulate media and AI-interaction logic.
• “server” folder contains index.ts (app entry), routes.ts (API layer), services/ (AI integration), db.ts (Drizzle setup), storage.ts, vite.ts (for hot-reload).
• Shared schema in “shared/schema.ts” centralizes database models.
• Config files for Drizzle, Vite, Tailwind, PostCSS, TS config reflect modern build pipeline.
README for additional insights
• README explicitly calls out a three-layer architecture: Frontend, Backend, Data & Storage.
• Mentions external AI services (Google Gemini), PostgreSQL on Neon, in-memory file uploads, session persistence, security and performance optimizations.