Founding Agent Experience Engineer

Role overview Build the runtime that carries out tasks delegated to consumer AI agents, from interpreting a request through negotiating with a vendor agent and reporting the result. The role combines agent infrastructure with user-facing interaction design, giving people ways to follow progress, review decisions, and step in when needed.

Responsibilities - Design and ship the end-to-end runtime that dispatches consumer tasks to vendor agents using A2A and MCP. - Create streaming interfaces, conversation transcripts, and approval flows that let users intervene during negotiations. - Instrument conversations, tracking measures such as token use, tool calls, latency, and refusal reasons, and feed results into evaluations. - Work with product partners to translate broad goals, such as lowering a bill, into structured executable tasks. - Establish engineering practices for the agent platform, including code review, testing, and on-call playbooks.

Requirements - At least five years building production software with TypeScript and React, using Next.js or a similar framework. - Practical experience shipping LLM features such as tool use, function calling, streaming, and structured output. - Comfort designing systems around models that may produce incorrect results, change behavior, or stall. - Strong product judgment and the ability to improve on an initial product specification. - Ability and interest to work remotely across overlapping US time zones.

Nice to have - Experience with A2A, MCP, or a multi-agent framework. - Work on a consumer product involving delegated access to real accounts or money. - Familiarity with Supabase, Postgres, and edge functions.