RelayFlow
Full Stack Developer & Solutions Architect
RelayFlow is an integration-focused platform that connects CRM, payments, email, and internal process automation through a resilient event pipeline. Incoming vendor payloads are normalized into a canonical contract, routed into workflow tasks, and executed with retries, dead-letter recovery, and replay tooling.
Control plane + worker architecture for external event workloads
Dead-lettered events can be rehydrated and replayed safely
Duplicate inbound events do not create duplicate side effects
Correlation IDs link ingest, route, execute, and outcome
The Problem
Teams often spend engineering time maintaining fragile integration glue where providers are inconsistent, payload schemas evolve, and duplicate webhooks trigger unsafe side effects.
The Solution
Built a FastAPI control plane with asynchronous workers for orchestration. Added signature verification, idempotency strategies, canonical event storage, dead-letter queues, and operator tooling for replay and incident triage.
Technical Decisions
Key architecture decisions and their outcomes
Canonical event schema before workflow routing
Provider payloads were inconsistent and evolved independently across integrations.
Normalized all inbound payloads into a shared IntegrationEvent contract before orchestration.
Routing logic became stable and provider-specific parsing remained isolated.
Dead-letter queue as a first-class recovery path
Third-party outages and schema mismatches were unavoidable in production-like scenarios.
Routed retry-exhausted events to a dead-letter queue with replay controls.
Operators can recover failed workflows without data loss or manual re-entry.
Engineering Details
- Webhook endpoints verify signatures and persist raw + normalized payloads
- Idempotency keys combine provider, event type, and external IDs
- Workflow dispatcher schedules retryable tasks with exponential backoff
- DLQ tooling supports replay after root cause remediation
- Operator UI exposes queue depth, retry rate, and replay outcomes
Key Highlights
- Webhook ingestion with provider signature verification
- Canonical event contracts for cross-provider normalization
- Idempotent routing and retry-safe workflow execution
- Dead-letter queue and replay tooling for failed events
- Operator console for event timeline and failure inspection
- Contract tests for provider schema drift detection
Tech Stack
Skills & Technologies
Related Articles
Real-Time Everything: Why We Stopped Polling and Never Went Back
Our trading dashboard polled every 5 seconds and users complained about stale data. We rebuilt on Convex with real-time subscriptions and the difference was not incremental — it was a different product.
AI in Production: Lessons From Shipping to Real Users
Our first AI feature hallucinated a refund policy that did not exist. A customer followed it. Here is what we learned about putting language models in front of real people.
Observability Across Five Production Systems
Monitoring, tracing, and instrumentation look different in every system we run. Here is what we actually measure, how we measure it, and what those measurements have caught — across Kubernetes infrastructure, integration pipelines, high-performance APIs, browser automation, and AI agents.