Skip to content

System_Design.md

Latest commit

 

History

History
61 lines (49 loc) · 1.33 KB

System_Design.md

File metadata and controls

61 lines (49 loc) · 1.33 KB
Raw

NL Task Automator - System Design Document

1. Overview

NL Task Automator is a smart assistant that processes Hindi-English mixed language commands to manage personal tasks like reminders, emails, meet and Google Calendar events.

2. Features

  • Natural Language Command Processing
  • Task Categorization (reminder, email, meeting)
  • Real-time intent extraction
  • Google Calendar + Meet Integration
  • Email Automation
  • WebSocket Notifications

3. System Architecture

A. Frontend (React)

  • Task Input UI
  • Real-time feedback
  • Dark/Light toggle

B. Backend (Node.js + Express)

  • API for NLP processing
  • Google API integration
  • Reminder scheduler
  • Email sender

C. NLP Pipeline

  • Tokenization
  • Language detection
  • Rule-based + LLM intent classification
  • Dispatcher to modules

D. Database (MongoDB)

Collections:

  • Users
  • Tasks
  • Reminders
  • Email Logs

E. Third-Party Integrations

  • Google Calendar API (OAuth 2.0)
  • Nodemailer
  • Socket.IO (for live updates)

4. Data Flow

  1. User enters a task
  2. Frontend sends to /api/task
  3. NLP pipeline extracts intent
  4. Appropriate module executes
  5. Response sent back to frontend

5. Deployment

  • Frontend: Vercel
  • Backend: Render
  • Database: MongoDB Atlas

6. Future Enhancements

  • Telegram/WhatsApp bot
  • Voice input
  • LLM fine-tuning for better intent classification