Learn. Practice. Feel prepared.
GenAI & LLM Engineering Objective Questions and Answers - Set 4
This GenAI & LLM engineering quiz contains carefully curated objective questions with correct answers and clear explanations. It is designed for developers and ML engineers to test your skills in building, optimizing, evaluating, and deploying LLM-powered applications. This is part 4 of the series.
Practice GenAI & LLM Engineering MCQs with Detailed Explanations
Answer at least 12 questions to submit and verify answers.
46 What is a key challenge in evaluating LLMs for reasoning tasks? Hard
Correct answer:
Lack of reliable automatic metrics
Reasoning quality is hard to measure automatically and often needs human evaluation.
47 Which method improves robustness against adversarial prompts? Hard
Correct answer:
Instruction hierarchies and content filters
Instruction hierarchies and filters help prevent adversarial overrides.
48 What is the main benefit of asynchronous inference in LLM APIs? Medium
Correct answer:
Better throughput and resource utilization
Async inference improves throughput and resource utilization under high load.
49 Which practice best supports continuous improvement of prompts? Medium
Correct answer:
A/B testing with offline evaluation sets
A/B testing with evaluation sets enables data-driven prompt iteration.
50 What is the main risk of exposing internal prompts to end users? Hard
Correct answer:
Security and policy leakage
Exposed prompts can leak internal policies or enable prompt injection attacks.
51 Which approach improves determinism in structured outputs? Medium
Correct answer:
Greedy decoding with schema validation
Greedy decoding combined with schema validation improves deterministic structured outputs.
52 What is the main drawback of relying solely on fine-tuning instead of RAG? Medium
Correct answer:
Model knowledge becomes stale
Fine-tuned models can become outdated without access to fresh external data.
53 Which strategy best supports cost control in peak traffic scenarios? Hard
Correct answer:
Autoscaling with request batching
Autoscaling and batching help manage load and control costs during traffic spikes.
54 What is the key challenge of multi-agent LLM systems? Hard
Correct answer:
Coordination and error propagation between agents
Errors can propagate across agents, and coordination becomes complex.
55 Which technique improves grounding when retrieved context is noisy? Hard
Correct answer:
Context filtering and reranking
Filtering and reranking improve the quality of context passed to the LLM.
56 What is the main benefit of token streaming to frontend apps? Medium
Correct answer:
Better user experience with progressive output
Progressive output improves perceived responsiveness for users.
57 Which failure mode occurs when the LLM follows retrieved context even if it is incorrect? Hard
Correct answer:
Grounded hallucination
Grounded hallucination occurs when incorrect context is treated as authoritative.
58 Which method helps ensure structured JSON outputs from LLMs? Medium
Correct answer:
Schema validation and constrained decoding
Constrained decoding with schema validation enforces valid structured outputs.
59 What is the main operational risk of long-running agent workflows? Hard
Correct answer:
Cost overruns and cascading failures
Long workflows can accumulate cost and errors across steps, leading to cascading failures.
60 Which approach best supports safe deployment of autonomous LLM agents? Hard
Correct answer:
Human-in-the-loop approvals for critical actions
Human-in-the-loop approvals reduce risk for critical or irreversible actions.