Technical interview questions and drills

20 results across 1 active filter

Page 1 of 1

An AI feature became slower and more expensive. How would you diagnose it?

Diagnoses an AI regression by decomposing request stages, configuration changes, token growth, retries, routing, tools, retrieval, caching, and provider behavior.

Artificial IntelligenceLarge Language Models

Comparison questionHard~12 min

Drills

An AI feature became slower and more expensive. How would you diagnose it?

Compare a hosted model API with a self-hosted or open-weight model.

Compare inference, training, fine-tuning, prompting, and RAG.

How do temperature and sampling settings affect model output?

How do you balance quality, latency, and cost in an AI feature?

How should an application handle context overflow and truncated output?

How should an application handle refusals, timeouts, rate limits, and malformed model responses?

How would you choose a model for a product feature?

How would you design a safe first LLM feature for an existing product?

How would you roll out a model or model-version change safely?

What are tokens, and why do they matter in an AI application?

What does hallucination mean in an AI application?

What goes into an LLM context window?

What is a large language model, and what does it actually predict?

What is an embedding, and how is it different from generated text?

What is the difference between model capability and application reliability?

When should you not use an LLM?

When would you use a smaller model instead of the most capable one?

Why can the same prompt produce different answers?

Why wrap a probabilistic model in deterministic application rules?