OpenAI Interview Questions

OpenAI's question bank is design-weighted. Of the 331 questions here, System Design accounts for 82 and ML System Design another 30, close behind Coding & Algorithms at 91; Machine Learning adds 28, Software Engineering Fundamentals 20 and Behavioral & Leadership only 13. 179 questions come from technical screens and 91 from onsite rounds, with small HR screen and online assessment sets. Software Engineer is the main track at 178 questions and Machine Learning Engineer at 77, then Data Scientist at 21 and a few Android, iOS and Frontend entries. The difficulty curve is steep: 160 medium, 117 hard and 10 easy. Payment infrastructure recurs across rounds, as authorization, capture and batch settlement, as a high-throughput authorization and settlement system, and as a full lifecycle from hold to charge. AI product design is the other cluster: an asynchronous text-to-video generation service, a large-scale streaming AI feature, command dispatch and telemetry reconciliation for unreliable devices. Fundamentals arrive as data structures with real semantics, such as a set with readable snapshots or serializing and restoring an in-memory key-value store, and the ML questions ask for implementation, such as vectorizing one-nearest-neighbor as a neural forward pass. 51 questions were added in the last 90 days, and 49 first-hand interview experiences are on the site.