1. What is Online inference ? Junior A. a Kubernetes framework for deploying and managing ML inference graphs and A/B tests B. a versioned store of trained models with stage tags (Staging, Production) and lineage C. serving low-latency predictions one request at a time behind a synchronous API D. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution
2. Which term means: "serving low-latency predictions one request at a time behind a synchronous API"? Junior A. Model registry B. Triton Inference Server C. Dynamic batching D. Online inference
3. Which statement is correct? Junior A. Online inference — NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. Online inference — a versioned store of trained models with stage tags (Staging, Production) and lineage C. Online inference — routing a small fraction of inference traffic to a new model version before full promotion D. Online inference — serving low-latency predictions one request at a time behind a synchronous API
4. What is Batch inference ? Junior A. a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints B. scoring a large dataset offline on a schedule rather than per request C. sending a copy of live traffic to a new model without serving its responses, to compare safely D. a versioned store of trained models with stage tags (Staging, Production) and lineage
5. Which term means: "scoring a large dataset offline on a schedule rather than per request"? Junior A. Model registry B. Online inference C. Batch inference D. Canary rollout
6. Which statement is correct? Junior A. Batch inference — a versioned store of trained models with stage tags (Staging, Production) and lineage B. Batch inference — scoring a large dataset offline on a schedule rather than per request C. Batch inference — serving low-latency predictions one request at a time behind a synchronous API D. Batch inference — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints
7. What is Model registry ? Mid A. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. a Kubernetes framework for deploying and managing ML inference graphs and A/B tests C. a versioned store of trained models with stage tags (Staging, Production) and lineage D. a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints
8. Which term means: "a versioned store of trained models with stage tags (Staging, Production) and lineage"? Mid A. Model registry B. Batch inference C. KServe D. Triton Inference Server
9. Which statement is correct? Mid A. Model registry — scoring a large dataset offline on a schedule rather than per request B. Model registry — a Kubernetes framework for deploying and managing ML inference graphs and A/B tests C. Model registry — a versioned store of trained models with stage tags (Staging, Production) and lineage D. Model registry — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints
10. What is KServe ? Mid A. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. grouping individual inference requests arriving close in time into one batch to raise GPU throughput C. a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints D. sending a copy of live traffic to a new model without serving its responses, to compare safely
11. Which term means: "a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints"? Mid A. Dynamic batching B. Batch inference C. KServe D. Model registry
12. Which statement is correct? Mid A. KServe — grouping individual inference requests arriving close in time into one batch to raise GPU throughput B. KServe — routing a small fraction of inference traffic to a new model version before full promotion C. KServe — a versioned store of trained models with stage tags (Staging, Production) and lineage D. KServe — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints
13. What is Seldon Core ? Mid A. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. sending a copy of live traffic to a new model without serving its responses, to compare safely C. scoring a large dataset offline on a schedule rather than per request D. a Kubernetes framework for deploying and managing ML inference graphs and A/B tests
14. Which term means: "a Kubernetes framework for deploying and managing ML inference graphs and A/B tests"? Mid A. Model registry B. KServe C. Seldon Core D. Shadow deployment
15. Which statement is correct? Mid A. Seldon Core — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints B. Seldon Core — a Kubernetes framework for deploying and managing ML inference graphs and A/B tests C. Seldon Core — grouping individual inference requests arriving close in time into one batch to raise GPU throughput D. Seldon Core — serving low-latency predictions one request at a time behind a synchronous API
16. What is Triton Inference Server ? Mid A. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints C. scoring a large dataset offline on a schedule rather than per request D. serving low-latency predictions one request at a time behind a synchronous API
17. Which term means: "NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution"? Mid A. Model registry B. KServe C. Online inference D. Triton Inference Server
18. Which statement is correct? Mid A. Triton Inference Server — scoring a large dataset offline on a schedule rather than per request B. Triton Inference Server — NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution C. Triton Inference Server — sending a copy of live traffic to a new model without serving its responses, to compare safely D. Triton Inference Server — a Kubernetes framework for deploying and managing ML inference graphs and A/B tests
19. What is Dynamic batching ? Senior A. routing a small fraction of inference traffic to a new model version before full promotion B. scoring a large dataset offline on a schedule rather than per request C. grouping individual inference requests arriving close in time into one batch to raise GPU throughput D. serving low-latency predictions one request at a time behind a synchronous API
20. Which term means: "grouping individual inference requests arriving close in time into one batch to raise GPU throughput"? Senior A. Model registry B. KServe C. Shadow deployment D. Dynamic batching
21. Which statement is correct? Senior A. Dynamic batching — grouping individual inference requests arriving close in time into one batch to raise GPU throughput B. Dynamic batching — sending a copy of live traffic to a new model without serving its responses, to compare safely C. Dynamic batching — a Kubernetes framework for deploying and managing ML inference graphs and A/B tests D. Dynamic batching — routing a small fraction of inference traffic to a new model version before full promotion
22. What is Canary rollout ? Senior A. NVIDIA's server that hosts models from many frameworks with GPU batching and concurrent execution B. serving low-latency predictions one request at a time behind a synchronous API C. routing a small fraction of inference traffic to a new model version before full promotion D. sending a copy of live traffic to a new model without serving its responses, to compare safely
23. Which term means: "routing a small fraction of inference traffic to a new model version before full promotion"? Senior A. KServe B. Canary rollout C. Online inference D. Seldon Core
24. Which statement is correct? Senior A. Canary rollout — routing a small fraction of inference traffic to a new model version before full promotion B. Canary rollout — sending a copy of live traffic to a new model without serving its responses, to compare safely C. Canary rollout — a Kubernetes framework for deploying and managing ML inference graphs and A/B tests D. Canary rollout — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints
25. What is Shadow deployment ? Senior A. a Kubernetes framework for deploying and managing ML inference graphs and A/B tests B. serving low-latency predictions one request at a time behind a synchronous API C. sending a copy of live traffic to a new model without serving its responses, to compare safely D. grouping individual inference requests arriving close in time into one batch to raise GPU throughput
26. Which term means: "sending a copy of live traffic to a new model without serving its responses, to compare safely"? Senior A. Shadow deployment B. KServe C. Triton Inference Server D. Dynamic batching
27. Which statement is correct? Senior A. Shadow deployment — grouping individual inference requests arriving close in time into one batch to raise GPU throughput B. Shadow deployment — sending a copy of live traffic to a new model without serving its responses, to compare safely C. Shadow deployment — a Kubernetes-native model-serving framework providing autoscaling InferenceService endpoints D. Shadow deployment — routing a small fraction of inference traffic to a new model version before full promotion