Model Serving
11 practice items · needed by 2 roles
Easy 5Medium 2Hard 4
Practice questions
- Why export a model to ONNXeasy
- Tuning an HNSW vector indexhard
- When batch inference fitseasy
- Traffic before the model has loadedeasy
- Why export a model to ONNXeasy
- Protocol for internal inference callseasy
- Low GPU use under loadmedium
- Silent wrong predictions from bad inputmedium
- p99 latency spikes during traffic burstshard
- Hundreds of low-traffic modelshard
- State leaking between requestshard