Anthropic MLOps Engineer Interview Questions

Company/scenario question on open vs closed LLMs — self-hosting economics, compliance, upgrade velocity, and hybrid strategies.

Advanced scenario question on why LLMs hallucinate — next-token prediction vs fact verification, root causes, and production defenses asked at OpenAI, Google, Meta, Anthropic, Microsoft, and Amazon.

Part 2 of the LLM hallucination handbook — factual, citation, reasoning, math, code, temporal, context, and tool-use hallucinations; decoding risks; RAG failure modes; production case study.

Part 3 of the hallucination handbook — production defense layers from retrieval and prompting through guardrails, citations, confidence scoring, eval frameworks, and enterprise architecture. Asked at OpenAI, Google, Meta, Anthropic, Microsoft, and Amazon.

Medium prompt engineering question on prompt unit tests and CI gates.

**Bedrock** AWS-native security, CloudTrail, existing EA — model versions may lag.

**Desktop host** Users add MCP servers in config; stdio local servers common.

**Mechanics** Assistant message may include tool_use blocks with id, name, input JSON; user message returns tool_result tied to id.

Deployment topology shapes MCP architecture — interviewers test practical product sense.

Tool schema drift breaks agents silently — models call old argument shapes and get 400s. Interviewers want a semver-minded process.

MCP auth interviews test whether you treat agents as privileged automation — because they are.

Build golden sets, error taxonomy, automated graders, and human review loops to iterate prompts like production software.

Evaluate persona and role prompts for tone, domain framing, and when they add noise without improving task accuracy.

Foundational prompt engineering question on instruction placement, example selection, and when demonstrations beat bare instructions.

Hard AI Agents interview question on browser agents and computer use risks — architecture, trade-offs, eval, and production patterns.

Hard RAG interview question on evaluating rag: faithfulness, context precision, recall — architecture, trade-offs, eval, and production patterns.

Scenario question on cross-model tokenizer bugs — chunk boundaries, context overflow, and embedder vs LLM alignment.

Model question on log probabilities — uncertainty signals, calibration limits, and confidence workflows.

Scenario question comparing CoT prompting with calculators, code interpreters, and search tools for reliable reasoning.

Scenario question on abstention — retrieval scores, calibration, UX of uncertainty, and avoiding confident wrong answers.

Hard scenario on jailbreaks — DAN, prompt injection, alignment limits, red-teaming, and layered defenses.

Hard project question on use-case benchmarking — golden sets, task metrics, regression gates, and leaderboard limitations.

Easy project question on SSE/token streaming — perceived latency, client backpressure, partial JSON, and disconnect recovery.

Project question on prefix/prompt caching and KV cache reuse — TTFT optimization and provider billing implications.