AI & LLMs

Agent Engineer

An engineer who builds, tunes and operates customer-specific agents, and owns their behaviour in production.

An agent engineer is a forward-deployed role with an agent-shaped mandate: tool schemas, escalation paths, guardrails, failure handling, and evaluating non-deterministic behaviour against a customer’s real workflows. Sierra popularised the title for exactly this work.

Worked example: a support agent that resolves 60% of tickets and cleanly escalates the rest is deployable; one that resolves 85% and confidently mishandles the remainder is not, because the failures surface in front of the customer’s customers. Gotcha: interviews for the role frequently ban agent frameworks, because building the loop by hand is the only way to show you understand termination conditions, tool-result formatting and context management.