How to Design Uber Eats: A System Design Mock Interview

This blog post summarizes a mock interview discussion on designing a system like Uber Eats. It explores the technical considerations involved in building a large-scale food delivery platform.

Key Requirements:

Restaurants can add, update, and remove themselves from the system.
Customers can view restaurants, search for them by various criteria (distance, delivery time), and place orders.
Delivery people receive notifications about new orders.

Non-Functional Requirements:

Scalability: The system must handle a large number of restaurants, customers, and orders.
Availability: The system should be highly available to ensure smooth operation.
Security: User data and financial information must be protected.
Latency: Response times for searching and viewing restaurants should be minimal.

Data Modeling:

Restaurants table: stores restaurant details like name, address, location (coordinates), and menu items.
Customers table: stores customer details like name, address, location (coordinates).
Menu items table: stores details about menu items like name, price, and image URLs.
Geohashing: A technique for dividing the world into grids to optimize location-based searches.

Database Selection:

Relational database with read replicas for restaurants (manageable size, frequent reads).
Sharded NoSQL database for customers (large user base, scalability).

System Design:

A central experience layer orchestrates services across different platforms (iOS, Android, web).
Restaurant service: manages restaurant data (adding, updating, removing).
Image service: handles image uploads and stores them in S3.
Image moderation API (optional): uses machine learning to moderate uploaded images.
Search service (Elasticsearch): facilitates searching for restaurants based on location and delivery time.
Caching layer: stores frequently accessed restaurant data to improve response times.

User Interaction Flow (Simplified):

Customer searches for restaurants using filters like location and delivery time.
Search service retrieves restaurants from Elasticsearch based on geohashes and delivery time estimates (isochrons).
Customer selects a restaurant and views details from the cache (if available) or restaurant service.
Customer places an order, which is routed to the relevant delivery person.

Conclusion:

This discussion provides a high-level overview of designing a complex system like Uber Eats. It highlights the importance of considering scalability, availability, security, and user experience when building such platforms.