How meamitpatil/zoom-learning-assistant works
Main Components
A. Client Layer:
• Zoom Chat Interface (Entry point for user commands)
• Zoom Webhook Endpoints
B. Application Layer:
• Flask Server
• Command Handlers
• Authentication Services
C. AI/Processing Layer:
• Embedding Processing
• Chat/LLM Processing (Fireworks AI)
• Vector Database (Qdrant)
D. External Services:
• Google Drive Integration
• Zoom API
• LlamaIndex
• Fireworks AI API
Component Relationships
[Zoom Chat] → [Zoom Webhook] → [Flask Server] → [Command Handlers]
• ↓
• [Authentication Services (zoom_auth.py)]
• ↓
• [Embedding Processing (embedding_utils.py)] ← → [Qdrant Vector DB]
• ↓
• [Chat Processing (chatgroq_handler.py)] ← → [Fireworks AI]
• ↓
• [Google Drive Upload (google_drive_upload.py)]
Important Details
A. Data Flow Annotations:
• Command inputs (/summarize, /start quiz, etc.)
• Webhook event handling
• Transcript processing flow
• Storage and retrieval
B. Key Integration Points:
• Zoom API authentication
• Vector database connections
• LLM API calls
• Google Drive upload process
Architectural Patterns to Highlight
• Event-Driven Architecture: Webhook handling for Zoom events
• Service-Oriented Architecture: Separate handlers for different functionalities
• API Gateway Pattern: Flask server managing various external service interactions
Implementation Details
• app.py (Main Flask application)
• zoom_webhook_handler.py (Webhook processing)
• chatgroq_handler.py (AI chat processing)
• embedding_utils.py (Vector embeddings)
• google_drive_upload.py (File storage)
• zoom_auth.py (Authentication)