Design a Text-to-Video Generation Service: GPU Scheduling, Failures, Traffic Bursts
Company: OpenAI
Role: Software Engineer
Category: System Design
Difficulty: hard
Interview Round: Onsite
Overview: Design the backend of a text-to-video generation service where each clip occupies GPUs for minutes: schedule jobs onto GPU workers, recover long-running jobs when workers fail, and absorb traffic bursts far beyond capacity. It tests asynchronous job design, GPU scheduling, checkpointing and admission control.