AI Engineer

Build the systems that let an AI colleague carry a task through tools, conversations, and human decisions. Success means the work gets done reliably, and people can understand what happened when it doesn’t.

• Build and evaluate agent workflows that use tools, retain relevant context, and ask for approval at the right moment.
• Turn real workflow failures into evaluation cases, then use them to compare models, prompts, and retrieval strategies.
• Improve recovery from interrupted sessions, failed tool calls, and incomplete information.
• Bring promising research into working product experiments, measuring reliability, latency, and cost before rollout.

• You have taken an LLM or ML system beyond a prototype and learned from how it behaved with real users.
• You can build data and evaluation pipelines and distinguish a convincing demo from a repeatable result.
• You are comfortable writing production software around models, including APIs, traces, and failure handling.
• You start with the user’s task and can explain when a simpler system is the better choice.

This is a preview role. Applications are not open yet.