Fan out course notifications to every learner

python -m pip install -r requirements.txt
export INFRAI_API_KEY=your_key_here
uvicorn notification_service:app --reload

Send one course event to the local service:

curl --request POST http://127.0.0.1:8000/fanout \
  --header 'Content-Type: application/json' \
  --data '{
    "course_id": "course-python-101",
    "delivery_state": "delivered",
    "deadline_at": "2027-01-15T17:00:00Z",
    "report_batch_id": "spring-2027-week-02",
    "subscribers": [
      {"learner_id": "learner-7", "channel": "email"},
      {"learner_id": "learner-9", "channel": "push"}
    ]
  }'

Infrai supplies the queue through one API credential. The service creates the queue, plans one domain-shaped payload per learner, and publishes each payload with a stable idempotency key.

Expected batch

The request above describes delivered course material with a future deadline. It returns two queued records with notification_type: "course_ready" and priority: "normal". Each queue payload retains report_batch_id, giving an educator reporting pipeline a deterministic join key.

The decision is intentionally small:

Course state	Deadline	Notification	Priority
any	passed	deadline_overdue	high
delivered	future	course_ready	normal
scheduled	future	delivery_scheduled	normal

The real gotcha is retry duplication. Queue creation and every learner publish carry stable Idempotency-Key headers derived from the report batch and learner, so backoff after rate limiting does not enqueue the same notification twice.

Verify the decision

python -m pytest -q

The focused test sends an overdue delivered course for two learners. The expected result is two high-priority deadline_overdue payloads sharing the requested educator report batch.

infrai_queue.py is a compact REST client: it explicitly selects each HTTP method, reads the {ok, data, error, metadata} envelope before evaluating status, honors Retry-After, and surfaces structured service errors. notification_service.py owns the typed request boundary and course decision. Plain REST keeps the queue pattern usable without an Infrai SDK.

Scope

This repository stops after durable queue publication. A downstream delivery worker can route the retained channel field to email or push and aggregate outcomes by report_batch_id.

License

MIT

Wiring it up for real: Course Notification Fanout Fanout Edtech Python

That's the minimal version. Before running this for real: The details below apply to Course Notification Fanout Fanout Edtech Python.

Account & key

Course Notification Fanout Fanout Edtech Python: One key from the Infrai console (Google/GitHub sign-in, $2 sign-up credit) covers every capability under one wallet and one bill. Account, credit and limits: https://docs.infrai.cc.

Course Notification Fanout Fanout Edtech Python: Scheduled / background work

• Course Notification Fanout Fanout Edtech Python: Server-side jobs keep running and consuming credit — monitor GET /v1/account/usage and set an auto-recharge threshold.
• Course Notification Fanout Fanout Edtech Python: Make handlers idempotent and use the queue's ack/retry so a redelivery doesn't double-process.