How d47r48yy6/starbridge_ai works
Project type and purpose
• This is a full-stack, server-side-rendered (SSR) web application built on Next.js.
• Purpose: An AI-Powered Interview Platform that generates role-specific questions, conducts voice interviews, analyzes responses, and stores feedback.
File structure
• app/ – Next.js App Router drives pages, layouts, and server actions.
• (auth) – Sign-in & sign-up flows
• (root)/interview – Interview listing, individual interview session, feedback pages
• api/vapi/generate – Serverless route for AI question/voice generation
• components/ui – Design system primitives (button, form, input, label)
• lib/actions – Business logic and abstractions for auth, general flows
• firebase/ – Firebase admin & client SDK wrappers
• public/ – Static assets (covers)
• constants/, types/, utils – Shared constants, TS types, utility functions
• next.config.ts, eslint.config.mjs, postcss.config.mjs – Build, lint, styling configs
README for additional insights
• Key services: Next.js SSR, Firebase Auth/Firestore, OpenAI GPT & Google Studio API (question gen), Vapi AI (voice interface)
• Architecture hints: Server-side routes for AI calls, Firebase-driven backend for user data & NLP evaluation
Main components & how to represent them
a. Frontend (Client Browser)
• Next.js pages & components
• React-based UI in components/ui
• CSS via globals.css + Tailwind/PostCSS
b. Server-Side (Next.js / Vercel)
• SSR compute and server actions in app/…/route.ts
• API Route: /api/vapi/generate → orchestrates GPT + Google Studio + Vapi AI
• Firebase client SDK for auth + Firestore queries
c. Backend Services
• Firebase Auth: user sign-up, sign-in
• Firebase Firestore: store interview sessions, responses, feedback
• Firebase Cloud Functions (optional for NLP evaluation) via firebase/admin.ts
d. External AI Services
• OpenAI GPT & Google Studio API: dynamic question generation
• Vapi AI: voice interaction, text-to-speech & speech-to-text
e. Data Flows
• User → Browser → Next.js SSR → API routes → AI Services → Response → Browser
• Browser → Firebase Auth → Firestore → Browser
• AI-generated feedback stored in Firestore → Retrieved on feedback pages
Architectural patterns & design principles
• Folder-based routing (Next.js App Router)
• Separation of concerns: UI components vs. business logic (lib/actions)
• Serverless functions for “backend” endpoints
• Single-source truth for types (types/) and constants (constants/)
• Design system in components/ui for consistency
Elements
• Clear labels: “Next.js SSR”, “API Route: /api/vapi/generate”, “Firestore”, “OpenAI GPT”, “Vapi AI”
• Blue for client-side layers
• Green for serverless/API layers
• Orange for external AI services
• Yellow for data stores (Firestore)
• Rectangles for compute layers (SSR, API, Cloud Functions)
• Clouds for external services