Agent Engineer
We believe the future of company operations is multi-agent systems — not copilots that assist a person through a task, but systems that carry work end to end with people supervising the edges. Most companies are still years from running this way.
We are building toward it now, and this role leads it.
§ 01 What You'd Own
The surface area.
- The architecture. How agents are orchestrated, how state and context move between them, how work is handed off, and where a human stays in the loop.
- Reliability and evaluation. Systems that run unattended need evals, regression tests and failure handling. Output that is wrong and confident is worse than output that is missing.
- The platform underneath. Tooling, MCP servers, integrations, permissions, cost and latency.
- The next thing to automate. Finding the work that recurs, and ending it. This role is judged by what stops being manual, not by what ships.
§ 02 What We're Looking For
Who this fits.
- You have built agentic systems that run in production without a human watching, and you have opinions about why most of them fail.
- Strong Python or TypeScript. Comfortable with orchestration, tool and function calling, structured output, retries and checkpointing.
- Real evaluation discipline. You measure whether the system is right, not whether it ran.
- Judgment about where autonomy belongs and where it doesn't.
§ 03 Helpful, Not Required
Nice to have.
- Extracting structure from messy, inconsistent source material.
- MCP server development.
- Experience in a regulated domain where a wrong answer has consequences.
§ 04 Compensation
What it pays.
$140,000 – $160,000 base, plus meaningful equity
Apply for Agent Engineer
A short form — a link to something you've built, and two questions that take about fifteen minutes.
Or write to careers@fddhub.com. We read every application.