Skip to content
GenAI Prep
Home
Questions
Practice
Listen
Search questions
⌘K
Python
Coding
FastAPI
GenAI and LLMs
RAG
Evaluation
Agents
MCP
Databases
System design
Security
AWS and Docker
Projects
HR round
Eval
High priority
Hallucination, evaluation and observability
Grounding, guardrails, RAGAS metrics, LLM-as-a-judge, LangSmith tracing.
All
Not practiced
Saved
Asked in interviews
0/25 practiced
Expand all
Listen
Hallucination mitigation
10
1
What is hallucination in LLMs?
2
Why does hallucination happen in RAG?
3
How do you reduce hallucinations?
4
How do you ensure answers are grounded in retrieved context?
5
What happens when the retriever returns no relevant documents?
6
How do you implement a similarity threshold?
7
How do you handle conflicting information in retrieved documents?
8
How do you prevent the model from answering outside the supplied context?
9
How do you handle prompt injection?
10
How do you validate generated responses?
GenAI evaluation frameworks
15
1
How do you evaluate a RAG application?
Diagram
2
Which evaluation framework are you using in your GenAI applications?
3
What is RAGAS?
4
Explain faithfulness.
5
Explain answer relevancy.
6
Explain context precision.
Diagram
7
Explain context recall.
Diagram
8
What is answer correctness?
Diagram
9
Difference between offline and online evaluation.
Diagram
10
How do you create a golden dataset?
Diagram
11
How do you evaluate retrieval independently from generation?
Diagram
12
What is LLM-as-a-judge?
Diagram
13
How do you evaluate latency and token consumption?
Diagram
14
What is LangSmith?
Diagram
15
How do you trace an agent's execution?
Diagram
Previous
RAG
Next
Agents