Design an Online Coding Judge
Company: Meta
Role: Machine Learning Engineer
Category: System Design
Difficulty: easy
Interview Round: Onsite
Design a large-scale online coding practice and interview platform similar to a programming challenge website.
The system should support:
- browsing coding problems
- viewing problem statements and sample test cases
- writing code in the browser
- running code against sample tests
- submitting code against hidden tests
- returning verdicts such as Accepted, Wrong Answer, Time Limit Exceeded, Runtime Error, and Compilation Error
- storing submission history
Assume the following scale:
- 20 million monthly active users
- 200,000 concurrent active coding sessions at peak
- 50,000 submissions per minute at peak
- support for 10 programming languages
- median submission verdict latency under 5 seconds
Discuss:
- APIs and core data model
- high-level architecture
- code execution and sandboxing
- queueing and scheduling of judge jobs
- storage for problems, test cases, code, and results
- scaling strategy
- security and isolation
- observability and failure handling
Quick Answer: This question evaluates system design and scalability skills for building a high-throughput, low-latency online coding judge, covering distributed architecture, queuing and scheduling, secure sandboxed code execution, data modeling for problems, submissions and results, and observability.