Sky Grills: A Real-Time Restaurant Operating System on the Edge
How we built a full restaurant platform: online ordering, staff POS, kitchen display, and live order tracking, on Cloudflare Workers with WebSocket realtime and Paystack payments.

Restaurants run on speed and coordination. The kitchen needs orders the moment they're placed. The counter needs to know when food is ready. The customer needs to see where their order is. Most restaurant systems handle one of these well; few handle all of them in real time, without a rack of on-premise servers.
Sky Grills came to us with exactly that ambition: a single platform that runs the whole restaurant (online storefront, staff point-of-sale, kitchen display, and live order tracking) all in real time. We built it on Cloudflare's edge.
"From the moment a customer taps order, to the plate hitting the counter, every step is tracked live. That's the Sky Grills experience we engineered."
The Edge Stack
- Cloudflare Workers + Hono: High-performance routing with sub-10ms response times.
- Durable Objects: Powers WebSocket real-time rooms connecting Kitchen Displays (KDS) and customer tracking pages.
- D1 + Drizzle ORM: Transactional storage with menu, order, and inventory state.
- KV + Queues: Idempotent payment processing, thermal receipt printer dispatch, and webhook retries.
Realtime FIFO Kitchen Workflow
The heart of the system is the WebSocket realtime hub built on a Durable Object. When an order is placed:
- It immediately appears on the Kitchen Display System (KDS).
- Chefs move orders through a FIFO queue (
START->COOKING->READY). - The customer tracking screen updates instantly over WebSocket without polling.
- Counter staff confirm pickup with barcode/PIN verification.
Resilient Offline-Tolerant Payments
Payment reliability is non-negotiable. Using Paystack with idempotency locks in KV, retried webhooks never cause double-charges. If network connectivity flickers in a physical kitchen, background Queues buffer commands until the link recovers.
DevBlock Engineering
Autonomous AI agents, fine-tuned models, and edge infrastructure engineered for enterprise production.