Build a geo-distributed Ad Serving Pipeline

Last updated: November 7, 2025

Quick Overview

Design a geo-distributed ad serving system that handles millions of requests. Discuss trade-offs in consistency, availability, and performance.

OpenAI
System Design
Machine Learning Engineer
OpenAI
November 7, 2025
Machine Learning Engineer
Technical Screen
System Design
Easy

126

3

112 solved


Design a geo-distributed ad serving system that handles millions of requests. Discuss trade-offs in consistency, availability, and performance.

System design interviews at OpenAI typically last 45-60 minutes. You are expected to drive the conversation, starting from requirements gathering through to a detailed architecture. The interviewer will evaluate your ability to handle ambiguity and make practical engineering decisions.

What the Interviewer Expects
  • Clearly define functional and non-functional requirements
  • Propose a reasonable high-level architecture with core components
  • Choose appropriate data storage solutions with basic justification
  • Discuss basic scaling strategies (horizontal scaling, caching)
  • Identify potential bottlenecks and suggest simple solutions
Key Topics to Cover
Message queues and async processing
Load balancing and horizontal scaling
High-level architecture and component design
Security and authentication
Consistency models and replication
Database selection and data modeling
How to Approach This
  1. Start by clarifying functional and non-functional requirements with the interviewer.
  2. Estimate the scale: QPS, storage, bandwidth. This drives your design decisions.
  3. Draw a high-level architecture first, then deep dive into 1-2 critical components.
  4. Discuss trade-offs explicitly (e.g., consistency vs availability, SQL vs NoSQL).
  5. Address failure scenarios, monitoring, and how the system handles 10x traffic spikes.
Possible Follow-up Questions
  • What happens if one of your database nodes goes down?
  • How do you ensure data consistency across multiple services?
  • How would you migrate from a monolithic to a microservices architecture?
  • How would you handle a region-wide outage?
Practice a Similar Problem on Codemia

Solve a related problem with our interactive workspace, get AI feedback, and view detailed solutions.

Solve on Codemia
Sample Answer
Requirements

Functional Requirements

  1. Ad Request Handling: The system must handle millions of ad requests per second from various geographical locations.
  2. Ad Selection Algorithm: Implement an ad se...
Capacity Estimation

Back-of-Envelope Calculations

  1. Traffic Estimation:
    • Assume 10 million daily active users, each making 10 ad requests per day.
    • Total requests = 10 million users * 10 requests = 100 m...

Submit Your Answer
Markdown supported

Related Questions