NCP-AAI 시험 - NVIDIA실제시험문제와 답 - 121문항

Question No : 1

When evaluating an agent’s degrading response times under increasing load, which analysis approach most effectively identifies scalability bottlenecks and optimization opportunities?

A.Track average response time while examining stage-by-stage processing metrics, resource usage trends, and potential components impacting scalability.
B.Test at fixed, low load levels while using controlled stress scenarios to compare with performance under production-like traffic patterns.
C.Profile each major system stage using distributed tracing, analyze GPU utilization with NVIDIA performance tools, and map queuing delays against varying workload patterns.
D.Focus on model inference duration while also measuring preprocessing time, tool-calling latency, and response formatting in the end-to-end pipeline.

정답:

Question No : 2

Which two deployment patterns are MOST suitable for scaling agentic workloads on NVIDIA Infrastructure? (Choose two.)

A.Bare metal deployment with manual resource allocation
B.Static virtual machine deployment with fixed resources
C.Serverless deployment without GPU acceleration
D.Containerized deployment with NIM (NVIDIA Inference Microservices)
E.Kubernetes orchestration with Horizontal Pod Autoscaling (HPA)

정답:

Question No : 3

You are deploying a multi-agent customer-support system on Kubernetes using NVIDIA GPU nodes and Triton Inference Server. Traffic spikes during product launches. You need <100ms response times, zero downtime, automatic GPU scaling, and full monitoring.
Which deployment setup best achieves cost-effective, reliable, low-latency scaling?

A.Set up one mixed GPU node pool with Cluster Autoscaler min=0, scale by network throughput, monitor via metrics-server and logs, and skip readiness probes for fast startup.
B.Place GPU pods on on-demand nodes in one zone, disable Cluster Autoscaler, run a fixed pod count for bursts, scale on CPU usage, and monitor with default health checks.
C.Deploy GPU pods in a node pool spanning all zones, mix GPU types, enable Cluster and Horizontal Pod Autoscalers using Prometheus GPU and latency metrics, and monitor with NVIDIA DCGM and Grafana.
D.Use spot-instance node pools across zones, enable Cluster Autoscaler with capped nodes, scale on memory usage, and monitor with logs and cluster events.

정답:

Question No : 4

A company is deploying a multi-agent AI system to handle large-scale customer interactions. They want to ensure the system is highly available, cost-effective, and scalable across multiple NVIDIA GPUs using container orchestration tools .
Which practice is most crucial for successfully deploying and scaling an agentic AI system in production?

A.Use a static assignment of requests across agents to maintain consistent agent operation and simplify coordination while scaling infrastructure resources as needed.
B.Optimize GPU utilization frameworks with workload optimization separate from cost analysis, prioritizing resource performance for peak load scenarios in deployment.
C.Deploy agents on a single machine to obtain a dimensioning baseline and thereby reduce setup complexity before expanding system scope.
D.Implementing automated workload management and resource scheduling frameworks to optimize GPU utilization and maintain service availability.

정답:

Question No : 5

A social media company wants to expand its agentic system to support global users, minimize downtime, and ensure smooth operation during usage spikes. The team is considering various deployment and scaling strategies to achieve these goals .
Which solution most effectively supports reliable and scalable deployment for an agentic AI system serving a global user base?

A.Integrating MLOps practices for continuous deployment and rapid model updates in production environments
B.Designing a distributed system architecture with multi-region deployment, automated failover, and dynamic resource allocation
C.Implementing containerization with Docker to simplify deployment and streamline updates
D.Using hardware profiling to optimize agent workloads for efficient GPU utilization across all deployed instances

정답:

Question No : 6

A company plans to launch a multi-agent system that must serve thousands of users simultaneously. The team needs to ensure the system remains reliable, scales efficiently as demand increases, and operates in a cost-effective manner.
Which approach is most effective for achieving robust and scalable deployment of an agentic AI system in production?

A.Running agents without load balancing to reduce infrastructure complexity and achieve robust and scalable deployment of an agentic system
B.Establishing a continuous monitoring framework to track system performance and adapt resources as usage patterns evolve
C.Deploying all agents on a single server with ongoing performance monitoring to maximize hardware utilization
D.Orchestrating agents using containerization platforms, combined with load balancing and ongoing performance monitoring

정답:

Question No : 7

What benefits does a Kubernetes deployment offer over Slurm?

A.Kubernetes provides autoscaling, auto-restarts, dynamic task scheduling, error isolation with containers, and integrated monitoring.
B.Kubernetes is the best option for both training and inference, offering advantages for resource management and workload visibility over traditional HPC schedulers like Slurm.
C.Kubernetes is more optimized for batch jobs to achieve high throughput, and also provides for monitoring and failover in large-scale workloads.

정답:

Question No : 8

When analyzing performance bottlenecks in a multi-modal agent processing customer support tickets with text, images, and voice inputs, which evaluation approach most effectively identifies optimization opportunities?

A.Measure total response time as this analyzes aggregated performance trends across modalities, model loading times, and opportunities for parallel execution.
B.Profile end-to-end latency across modalities, measure model switching overhead, analyze batch processing opportunities, and evaluate Triton’s dynamic batching for multi-modal workloads.
C.Optimize each modality independently using dedicated profiling of cross-modal interactions, shared resource constraints, and pipeline execution strategies.
D.Extend evaluation to accuracy and quality metrics, incorporating resource usage patterns, latency observations, and their impact on user experience.

정답:

Question No : 9

When analyzing throughput bottlenecks in a multi-modal agent processing text, images, and audio, which Triton configuration evaluations identify optimization opportunities? (Choose two.)

A.Analyze model ensemble pipelines for sequential dependencies, identify parallelization opportunities, and optimize inter-model data transfer using Triton’s scheduler.
B.Profile GPU memory allocation patterns across modalities, implement model instance batching strategies, and tune concurrency limits to maximize utilization.
C.Deploy each modality on separate Triton instances, allowing Triton to automatically manage ensemble coordination, shared memory usage, and pipeline integration.
D.Use a single model instance per GPU, allowing Triton to automatically optimize concurrency, batching, and multi-instance settings for throughput scaling.

정답:

Question No : 10

An e-commerce platform is implementing an AI-powered customer support system that handles inquiries ranging from simple FAQ responses to complex product recommendations and technical troubleshooting. The system experiences unpredictable traffic patterns with sudden spikes during sales events and varying complexity requirements. Simple questions comprise the majority of requests but require minimal compute, while complex product recommendations need sophisticated reasoning. The company wants to optimize costs while maintaining service quality across all query types.
Which approach would provide the MOST cost-optimized scaling strategy for this variable-workload, mixed-complexity environment?

A.Deploy specialized NVIDIA NIM microservices using a single large model configuration that handles all agent functions on high-capacity GPUs, with auto-scaling infrastructure that maintains constant resource allocation across all traffic patterns.
B.Deploy specialized NVIDIA NIM microservices on CPU-optimized infrastructure with auto-scaling capabilities to minimize hardware costs, while accepting longer inference times for cost optimization benefits.
C.Deploy specialized NVIDIA NIM microservices with an LLM router to dynamically route requests to appropriate models based on complexity, combined with auto-scaling infrastructure that scales different model types independently.
D.Deploy multiple specialized NVIDIA NIM microservices with identical high-capacity models across all available GPUs, implementing auto-scaling infrastructure without request complexity differentiation or dynamic model selection capabilities.

정답:

Question No : 11

You’re utilizing an LLM to translate complex technical documentation into multiple languages. The translations often lack nuance and fail to capture the original intent.
What’s the most effective strategy for improving the quality of the translations?

A.Providing the LLM with a glossary of key terms, concepts in all languages and the dataset of previously translated text.
B.Training the LLM on a dataset of translated texts.
C.Providing the LLM with guidance to “translate the documents” without additional guidance, so it can use trained knowledge.
D.Providing the LLM with guidance to translate “with high accuracy” without additional guidance, so it can use trained knowledge.

정답:

Question No : 12

Your agent is generating inconsistent and contradictory statements .
Which approach would be most suitable to improve the agent’s output?

A.Employing Reflexion
B.Increasing the number of generated plans
C.Using Decomposition-First Planning
D.Decreasing the length of prompts

정답:

Question No : 13

You are using an LLM-as-a-Judge to evaluate a RAG pipeline.
What is the primary benefit of synthetically generating question-answer pairs, rather than relying solely on human-created test cases?

A.Synthetically generated questions are more challenging and reveal deeper flaws in the RAG pipeline.
B.Synthetic generation eliminates the need for any human validation of the RAG pipeline’s output.
C.Synthetically generated answers are inherently more accurate than those produced by the LL
D.Synthetic generation allows for systematic testing of the RAG pipeline across a wider range of scenarios and query types.

정답:

Question No : 14

When analyzing inconsistent performance across a fleet of customer service agents handling similar queries, which evaluation approach most effectively identifies root causes and optimization opportunities?

A.Assess performance data from recently improved agents and highlight strong results, using outcome comparisons to identify areas with the greatest impact on service quality.
B.Average performance metrics across all agents as this will smooth individual variations, query distribution differences, and temporal factors affecting agent behavior and accuracy.
C.Deploy stratified evaluation sampling across agent variants, query complexity levels, and temporal patterns while tracking decision paths using comparative analytics.
D.Review performance across both high- and low-accuracy agent groups, comparing case outcomes and identifying patterns contributing to top and bottom results.

정답:

Question No : 15

A team is evaluating multiple versions of an AI agent designed for customer support. They want to identify which version completes tasks more efficiently, responds accurately, and improves over time using user feedback.
Which practice is most important to ensure continuous refinement and optimal performance of the AI agent?

A.Comparing agents on isolated tasks without standardized benchmarking pipelines
B.Relying solely on offline benchmarks without incorporating live user feedback during tuning
C.Implementing an evaluation framework that quantifies task efficiency and incorporates human-in-the-loop feedback
D.Tuning model parameters once before deployment to maximize initial accuracy

정답:

NVIDIA NCP-AAI 시험