You will build a Contact Center as a Service (CCaaS) system that processes call data from external providers using an event-driven architecture. The system receives webhook notifications when calls are completed, processes them asynchronously using AI, and sends results back via callbacks.

• FastAPI web server with webhook endpoints
• FastStream + Redis for event-driven message processing
• AI Processing service for call transcription and summarization
• Async workers for background processing

I let you discover how the project is made to do the setup correctly

I let you discover the project and then we can talk a bit more about the system architecture

File: app/services/ai_processor.py

Implement the AIProcessor.process_call() method to:

1. Download audio from recording_url
2. Transcribe audio to text
3. Generate summary from transcript
4. Send results to callback_url as POST request

Key considerations:

• It is important to have flexibility on how we want to do transcription and summarization (calling external API or internal API for example)
• Handle errors gracefully
• Make the system easily testable

File: tests/test_e2e.py

Test the complete flow from webhook to callback:

Note: This requires Redis to be running. Consider using test fixtures for broker setup.

Send test webhook:

curl -X POST http://localhost:8000/webhooks/call-completed \
  -H "Content-Type: application/json" \
  -d '{
    "call_id": "test-123",
    "provider": "twilio",
    "recording_url": "https://httpbin.org/delay/1",
    "callback_url": "https://httpbin.org/post"
  }'

During the session, we'll discuss:

• Design patterns for external service integration
• Testing strategies for async/event-driven systems
• Error handling and resilience patterns
• Scaling considerations for AI processing workloads

Good luck! 🚀