This document explains the high-level architecture of the Intervyo platform.
User (Browser) → Frontend (React) → Backend API (Node.js + Express) → AI Evaluation Engine → MongoDB Database → Analytics & Feedback Layer
- React with Tailwind CSS
- Handles interview UI, timers, dashboards
- Communicates with backend via REST APIs
- Uses browser APIs for speech and media input
- Node.js with Express
- Handles authentication, interviews, evaluation, analytics
- Orchestrates AI interactions
- Stores and retrieves data from MongoDB
- Uses structured prompts
- Scores responses using fixed rubrics
- Generates follow-up questions
- Identifies strengths and weaknesses
- Predicts future failure points
Core collections:
- Users
- Interviews
- Evaluations
- Questions
- Analytics
- AttackPlans
- CalendarEvents
- JWT-based authentication
- Role-based access control
- Private interview data by default
- User starts interview
- Backend initializes session
- AI generates question
- User responds
- AI evaluates response
- Scores and feedback stored
- Next question triggered
- Simplicity over overengineering
- Deterministic evaluation over randomness
- Extensibility without breaking core logic
Architecture exists to support learning, not complexity.