Design a Food Delivery App

Problem Design a food delivery platform (DoorDash/Uber Eats style): browse restaurants and menus, place orders, match drivers, and track delivery in real time.

Functional requirements

• Restaurant/menu browsing with search and geo-filtering; cart and order placement with payment.
• Order lifecycle: placed → accepted by restaurant → prepared → picked up → delivered (with cancellations/refunds).
• Driver matching and live location tracking for the customer.

Non-functional requirements

• Scale to discuss: ~10M DAU, ~500 orders/sec peak (meal-time spikes), 100K concurrent drivers streaming location every few seconds.
• Order placement must be reliable/consistent; tracking can be eventually consistent but fresh (~2–5 s).

Key components

• Catalog service (restaurants/menus, search index, cache), order service (state machine, transactional store), payment service, dispatch service (driver matching), location service (driver GPS ingest — write-heavy time-series/redis geo), notification service (push/SMS), real-time tracking channel (WebSocket/SSE).

Deep dives / trade-offs

• Driver matching: nearest-driver via geo-index (geohash/H3 buckets); batching vs immediate assignment; handling rejections with timeout-and-reassign.
• Order state machine consistency across services — outbox/event-driven updates vs synchronous calls; exactly-once payment capture.
• Location ingest at 100K writes/sec: partitioning, in-memory geo store with periodic snapshot, fan-out to interested watchers only.