Chidi Onyema · Principal Backend Engineer
I build backend systems
that don't break
under pressure.
Payments that never double-charge. Orders that roll back cleanly. Services that stay up when dependencies fail. This is a live platform, not a slide deck.
Every demo sends real requests through 8 microservices running on Fly.io. The response times and trace IDs come from actual production services.
Real problems, real solutions
The hard parts of distributed systems.
Each card is a real failure mode I've solved. Click through to see it happen live.
A customer is charged but their order never appears.
The system detects the partial failure and automatically reverses the charge, releases the reserved stock, and cancels the order. No support ticket needed.
A customer panic-clicks 'Pay' three times during a slow connection.
They're charged exactly once. The second and third clicks are detected as duplicates and return the original receipt. Zero revenue leakage.
The payment provider goes down during Black Friday.
Traffic to the failing service is cut off in milliseconds. Browsing, search, and cart keep working. When the provider recovers, traffic resumes automatically.
Finance asks: where did this refund go, and who approved it?
Every pound in, every pound out. A complete audit trail where debits always equal credits. Money cannot be created, lost, or hidden.
Watch it work
What happens when checkout fails halfway through?
A checkout touches three services: stock, payments, and orders. Toggle "Payment fails" to see the system automatically undo everything that already happened. No stuck orders, no ghost charges.
The real demo sends requests through Cloudflare → BFF → RabbitMQ → Postgres.
Full interactive demo →Architecture
Patterns for scale and reliability.
Detailed breakdowns of the decisions that define the platform. Focusing on how we handle distributed state, data consistency, and failure at scale.
Saga vs Two-Phase Commit
When you can have ACID across services, when you cannot, and why the answer in 2026 is almost always sagas.
The Transactional Outbox
Why dual-write is the most expensive bug in event-driven systems, and the small amount of plumbing that fixes it forever.
Zero-Downtime Secret Rotation
How we rotate database credentials and API keys in a live distributed system without dropping a single request.