Direct-to-consumer retailer6 weeks2025Sample

Removing the CSV that ran the business

Order, inventory, and fulfilment systems connected with durable pipelines — and alerting that notices when a sync stops.

This is an illustrative project archetype showing how we scope and describe work of this kind — not a delivered client engagement. Outcome metrics are withheld until this entry is replaced with a real project.

The challenge

Every morning someone exported orders, reformatted them, and uploaded the file to the fulfilment provider. When the export silently broke over a bank holiday weekend, nobody noticed for four days — the downstream report still arrived, reporting zero orders, exactly as designed.

What we did
  • Process mapping to identify which manual steps should never have been human
  • Idempotent, durable pipelines with classified retries and exponential backoff
  • Volume anomaly alerting: expected executions per period with a tolerance band
  • Dead letter queue with full failure context routed to a named owner
  • Manual override that does not require a deploy
Outcome

The daily export is gone. When a sync does fail it escalates within minutes to a named owner instead of surfacing days later.

Stack

PythonFastAPIVercel WorkflowStripeWebhooks