System Design · OpenAI · Hard
Design a large-scale near-duplicate detection service for visual media. Given an image or video upload, the service must identify previously stored assets that are perceptually equivalent even when the new file is a re-upload, a lightly edited version, a different encoding, or contains a watermark. The system should use detected matches to support takedown workflows, merging of duplicate content, and removal of redundant copies. The solution must cover the full pipeline: A…
Checking your access…