Design a GPU scheduling system that allocates a shared cluster across competing training and inference workloads.
A reported OpenAI design prompt. The skeleton: gang scheduling, topology-aware placement, preemption with checkpoint contracts, and quota, plus the fragmentation problem that quietly eats 20% of naive clusters.
Updated Aug 2026 · Grounded in real Forward Deployed Engineer interview loops and written to a senior-engineer editorial bar.
A reported OpenAI design prompt. The skeleton: gang scheduling, topology-aware placement, preemption with checkpoint contracts, and quota, plus the fragmentation problem that quietly eats 20% of naive clusters.
Lead with where the obvious approach breaks, because that is the judgment they are screening for — most candidates jump straight to the happy path and lose the room.
Then walk the failure back through the pipeline in order, naming the one metric the customer's exec sponsor actually cares about before you propose the fix.
