Victor Dantas

Distributed Systems / Payments

Distributed Payment Flow

A resilient payment workflow designed to reduce inconsistencies between orders, payments, refunds, and external providers.

Saga Pattern Idempotency Redis Payment Reliability External Providers Retry Strategy

Problem and context

Transactional consistency in a high-concurrency payment environment.

Payment flows often cross order services, external providers, refunds, notifications, and customer-facing states. The work focused on keeping those boundaries consistent when requests arrive concurrently or providers respond late.

Technical responsibilities

Designed backend flow boundaries, idempotency behavior, retry paths, Redis-backed locks, and failure handling around payment, reservation, order, and notification workflows.

Architecture decisions

Used saga-style orchestration, explicit state transitions, idempotent webhooks, delayed retries, DLQ handling, and Redis coordination to reduce duplicate processing and inconsistent state.

Result

A clearer payment architecture with better reliability around provider callbacks, concurrency control, and recovery from partial failures.