Interview Preparation

Interview Question Library

Browse practical engineering interview questions by role, topic, difficulty, and interview type. Open each question for a real-world sample answer, expert thinking, common mistakes, and related learning.

Question Library

Learn how strong engineers actually answer interview questions.

Every question is a Markdown content item with a real-world sample answer, expert guidance, follow-ups, common mistakes, and related learning. Browse, learn, then practice.

Questions

Roles

A model gives confident answers that are not supported by retrieved documents. How would you troubleshoot it?

RAGhallucinationgroundingevaluation

Interviewer is testing

• Layered diagnosis
• Evidence-based evaluation
• Fallback design

Learn how strong engineers actually answer interview questions.

A model gives confident answers that are not supported by retrieved documents. How would you troubleshoot it?

A model's production quality dropped even though the training pipeline did not change. What would you investigate?

A production API is intermittently timing out. Explain how you would isolate the failure and prevent recurrence.

A production API suddenly becomes slow. Walk through the first five things you would check before changing anything.

A production API's latency doubled overnight. What do you investigate first?

A RAG application has good latency but poor answer quality. How do you isolate the problem?

A RAG application has good latency but poor answer quality. How do you isolate the problem?

An event consumer is falling behind during traffic spikes. Design the recovery and scaling strategy.

An event consumer is falling behind during traffic spikes. How would you recover safely?

Design a multi-region API for a business that has explicit RTO and RPO requirements.

Design a production AI agent that can call business tools without becoming an uncontrolled automation risk.

Design a production ML platform that can safely promote models from training to serving.

Design a production RAG platform for an enterprise knowledge base with strict access controls.

Design a production RAG platform handling 10M queries per month.

Design a secure large-file upload flow without proxying every byte through the API.

Design an event-driven AWS pipeline that remains reliable when downstream processing slows down.

How do RTO and RPO change a multi-region architecture?

How do you balance cost against reliability when presenting an architecture to a customer?

How do you decide between a relational database and a NoSQL database for a new system?

How do you recognize when an architecture is becoming over-engineered?

How would you design a customer-facing application that survives a single availability-zone failure?

How would you design a data pipeline that can recover safely after a downstream processing failure?

How would you design least-privilege access for multiple application roles?

How would you explain a complex architecture trade-off to a non-technical stakeholder?

How would you keep an AI agent from performing an unsafe tool action?

How would you protect an API from a sudden 20× traffic spike while keeping legitimate users served?

How would you reduce the cost of a high-volume LLM application without sacrificing the required quality?

How would you safely roll out a new version of a Kubernetes workload?

Kubernetes workloads are being OOMKilled intermittently. How would you investigate?

Pick the most important project on your resume. What was the hardest engineering decision you made and why?

Reduce the cost of a high-volume LLM application without degrading its required quality target.

Tell me about a production incident you handled. What did you do and what changed afterward?

Tell me about a production incident you were involved in. Explain what happened, what you did, and what changed afterward.

Tell me about a technical decision where you disagreed with another engineer or stakeholder. How did you handle it?

Tell me about a time you improved an engineering system's cost, reliability, or delivery speed. What did you change and how did you measure success?

What does a safe CI/CD pipeline need beyond build and deployment steps?

What gates should a model pass before it is promoted to production?

What is RAG primarily used for?

What questions should you ask before designing a system from an ambiguous interview prompt?

What security controls would you expect around a production API?

What would you instrument in a distributed service before calling it production-ready?

When is event-driven architecture a better fit than direct service-to-service calls?

When should an API call stay synchronous, and when should work become asynchronous?

When would you accept weaker consistency to improve latency or availability?

When would you choose a managed service over more infrastructure control?

When would you introduce caching, and how would you prevent stale or incorrect data from becoming a problem?

Which statement best describes agentic AI?

Which two components commonly appear in an AI agent workflow?

Which two practices improve the quality of a knowledge-based AI application?

You need to process a task that may take several seconds. Explain when you would keep the request synchronous and when you would make it asynchronous.