Agent Engineer

What you'll do

This is not a job in the usual sense. It is a paid expert conversation: Mercor wants to hear, in detail, how engineers who run LLM agents in production actually do it. The ad does not name the client or say how the conversations will be used, though the topic list reads like research into agent tooling.

There are two steps:

1. A short conversational AI interview. No coding and no take-home. You talk through how you think about agent reliability, evaluation and adoption, and the tradeoffs you have made. The ad asks for war stories, "the messier and more specific, the better."
2. A live 30-minute call with the team, only if your screen stands out. This is the paid part.

Who fits

The ad describes the person it wants by experience rather than credentials. You have probably:

• Shipped an agent that real users depended on, and been responsible when it broke
• Worked out how to tell whether an agent got better or worse after a change
• Run agents that outlive a single request: scheduled jobs, long-running work, cloud sandboxes
• Watched your organisation roll out an internal assistant and seen who adopted it and who quietly did not

It is especially interested in the less-discussed layers: internal agents wired into company data, shared company memory, reusable skills and playbooks, the tool and MCP surfaces agents call, and observability of what agents did and what they cost.

If your agent work is prototypes and demos, you will struggle to supply the specifics this screen is fishing for. If you have run an evaluation pipeline for an agent in production, lead with that.

What it pays

$100 to $500 for the 30-minute live call, paid on completion, with the amount set by depth of experience. The AI screen itself is unpaid, and not everyone who completes it is invited to the call.

At the top of the range that is an excellent rate for half an hour. The ad does not mention follow-up sessions, so treat it as a one-off. Payment method, schedule and any visa or location restrictions are not published on this listing.

Worth knowing

Good:

• No coding test or take-home
• High effective rate for a single conversation
• Paid on completion of the call
• No location restriction is published

Less good:

• One-off; this is not ongoing income
• The screening interview is unpaid and selective
• Pay depends on a judgment of your depth, with a 5x spread
• Be careful about sharing confidential details of your employer's systems; talk about patterns, not proprietary specifics

About this listing

Posted by Mercor as a remote contract, confirmed open on 24 September 2026. The two-step process and the $100–500 range are the ad's own. See Mercor.