FDE interview questions by topic
Green questions are free. The full Evals section is free, plus the first 5 in every other topic.
Problem decomposition & customer discovery
The FDE signature round: ambiguous customer problem, scope it, ask discovery questions, pick a first deliverable, handle pushback.
Evals & measuring LLM quality
Designing eval sets, LLM-as-judge and its calibration, offline vs online metrics, regression testing, golden sets, failure taxonomies.
RAG & retrieval
Chunking, embeddings, hybrid search, reranking, citations, retrieval evals, freshness, permissions on customer data.
Agents, tool use & MCP
Tool calling, agent loops, MCP servers, guardrails, human-in-the-loop, cost and latency control, failure recovery.
LLM fundamentals & prompting
Tokens, context windows, sampling, prompt structure, structured output, caching, model selection, fine-tuning vs prompting.
AI system design for customer deployments
Designing an end-to-end AI system inside a customer's environment: data access, security, scaling, monitoring, rollout.
Practical coding & data (Python + SQL)
The live-coding and data rounds FDEs face: API integration, parsing messy data, SQL analytics, debugging scripts.
Debugging AI in production
Incidents: hallucination spikes, latency, rate limits, prompt injection, bad outputs, regressions after a model change.
Behavioral & ownership
Customer-facing ownership stories, conflict with customers or product teams, shipping fast, saying no, travel and ambiguity.
Company loops (who hires FDEs and how)
What is publicly reported about FDE / deployed-engineer loops at companies that actually hire for the role (for example Palantir, OpenAI, Anthropic, Databricks, Scale AI, Salesforce, Google, AWS). Only companies with public FDE-type roles.