Mercor's $150/hr Maintainers Aren't Moonlighting. They're Between Jobs

Mercor's job board is bidding for the engineers on your shortlist. It lists "Open Source Maintainer / Core Contributor" at $150/hr, and "Senior Software Engineering Experts" from top tech and quant firms at $120 to $180/hr. On Nov 10, 2026, Mercor announced a $750M Series D at a $20B valuation, double the $10B it reached in its October 2025 Series C. The money goes to evaluation and post-training work, so it can keep bidding for the pool you source on GitHub. The fine print on those listings helps you, though. It tells you when these engineers are reachable and what your pitch needs to beat hourly cash.

What do Mercor expert pay rates really look like?

$150/hr is the top of the range. On Nov 12, Mercor's Explore page listed backend software engineer work at $50 to $80/hr, and its homepage showed a $90/hr average contracted rate.

An AI training contractor is an expert paid by the hour to write, review or grade the data that teaches or tests a model. Mercor competes with several platforms for these people, and the posted software rates are close to each other:

The two listings in the news sit at the top of that band. Annualize them and you get the number your candidate has in mind when your offer arrives:

• Maintainer role: $150/hr × 40 hrs × 50 weeks = $300,000 gross.
• Senior SWE role: $120 to $180/hr × 2,000 hrs = $240,000 to $360,000 gross.

Those figures assume the work never stops, and it often does (more on that below). Your candidate may still anchor on them, so you need a pitch that competes on something other than the hourly rate.

Is a $150/hr Mercor contract a side gig?

No. The maintainer listing requires 30 to 40 hours a week, says the role "isn't a fit alongside a full-time job," and forbids work with any other AI lab or AI training company, including other Mercor projects, for the length of the engagement.

That changes the usual thinking about sourcing senior engineers with side gigs. The usual worry is that a well-paid side contract makes someone comfortable and hard to move. That applies to someone who does five hours a week of labeling at night. It doesn't apply here. A contract that takes up a full work week and requires exclusivity can only be taken by someone who has no full-time employer.

Think about who that is. Engineers rarely leave a FAANG-tier salary for an exclusive contract that can end early. More likely, the person who takes it is already between roles, and the contract pays the bills until the next job.

A maintainer billing Mercor 40 hours a week has already left their last job. That makes them reachable, not lost.

The senior SWE listing targets your hiring market almost exactly. It asks for 4+ years of professional experience at FAANG+, Mag 7+ or top quant firms, and it's open only to candidates in the United States, United Kingdom, Canada, Australia and New Zealand. If you hire senior engineers in any of those countries, these listings are recruiting from your pool.

Why does searching "Mercor" on LinkedIn find almost none of them?

Searching by Mercor or by trainer titles finds few of its senior contractors, most likely because they keep their Staff or Principal title and leave contract work off their profiles. In Refolk's index of professional profiles, 3,484 US profiles hold AI-trainer-type titles (AI Trainer, AI Training Specialist, AI Code Reviewer, Software Engineering Expert, AI Tutor), and Mercor does not appear among the top employers in a 25-profile sample.

The UK sample looks the same. Outlier leads with 5 of 25 profiles, followed by micro1 and DataAnnotation, and Mercor shows up once. In the US sample, Alignerr, Outlier and Micro1 lead. Both samples are small, but they agree.

Two things likely explain the gap:

1. Most of the "network" isn't doing the work. Mercor says it has more than 5 million experts. The same coverage describes more than 30,000 vetted experts, about 0.6% of the headline figure.
2. The vetted senior people keep their real title. A Staff Engineer with years at Google gains little by adding an hourly eval contract under that tenure, so it tends to stay off the profile.

So when you search AI training contractors on LinkedIn by title, you mostly find generalists at lower-paying platforms. The $150/hr people are more likely still listed as Staff or Principal Engineer, sometimes with a recent end date and nothing after it. Searching by title gets you the wrong people, so describe the person you want instead: what they maintain, where they worked, what they review. Refolk is built for that kind of query. You describe the engineer in plain English, and it searches GitHub, LinkedIn and the open web so the commit history and the employer history appear on one profile.

How do you find the maintainers Mercor is screening for?

Use Mercor's own filter. Its maintainer listing accepts anyone named in a MAINTAINERS or CODEOWNERS file, anyone with 10+ PR reviews in the last 12 months, and core contributors with 3+ substantive merged PRs. Every one of those signals is public on GitHub.

A CODEOWNERS file is a GitHub file that names who must review changes to specific paths in a repository. That makes it a public list of the people a project trusts. A MAINTAINERS file does the same job less formally.

234

Job titles won't find these people. Only 234 people across four countries call themselves maintainers in their headline. Even then, the title usually reads Principal Engineer or Staff Engineer, and "maintainer" sits further down. The index sample includes a "Lead Maintainer" at AboutCode/ScanCode, but that's rare. Most maintainers appear in repo files, not on profiles. Finding them by hand means reading CODEOWNERS files across dozens of repos, checking each name's review history, matching the GitHub handle to a real name and current employer, and then finding a way to contact them, for every role you open.

If you're doing the search yourself, this order works:

1. List the repos your stack actually depends on. Your lockfile beats any "top projects" list. A maintainer of a library you ship will understand your product on the first call.
2. Pull every name from MAINTAINERS and CODEOWNERS. These are the people Mercor would accept with no other evidence.
3. Filter by review activity. 10+ PR reviews in the last 12 months means they're still active, not just named in a file years ago.
4. Filter by merged PRs. 3+ substantive merged PRs identifies core contributors who aren't formally named.
5. Match to LinkedIn and look for the gap. A recent employer end date with nothing after it, or a vague "Independent" line, is the likeliest sign of someone working on a full-time contract.
6. Check geography. Mercor's senior listing covers only the US, UK, Canada, Australia and New Zealand, so candidates there face the most hourly-pay competition.

Steps 2 through 5 are where Refolk saves the most time. You can write the whole filter as one sentence:

How do you write a pitch that beats $150/hr?

Sell certainty, not a matching paycheck. The $300,000 annualized figure has no benefits and no equity, and Mercor's own payment FAQ says projects can be extended, shortened or ended early.

You will rarely match the hourly rate, and you don't need to. An hourly eval contract trades time for cash and leaves nothing behind. The candidate builds no equity, no public record, and no code with their name on it. Maintainers care about that record more than most engineers, because their reputation is their commit history. Compete on what the contract can't give them:

• A guaranteed run. A salaried role with a written start date and a real runway, against a project that can be shortened or ended early.
• Equity. Contract hours never turn into ownership. Even a modest grant is something the contract can't offer.
• Code they keep. Offer protected time to keep maintaining their project, with public commits. Eval work is confidential and nothing ships under their name.
• Benefits. Health coverage and retirement matching cost a contractor real money out of that $150.

Maintainers also have a good argument on their side. When the open-source LiteLLM project was compromised in March 2026, Mercor was among the firms hit. Companies depend on open-source maintenance, and a role that pays someone to keep doing it beats one that asks them to stop.

A short first message that applies all of this:

I saw you're in CODEOWNERS for [project] and reviewed most of the [module] changes this year. My team runs it in production. The role is a salaried Staff position with a written start date, equity, and one day a week protected for [project]. Happy to share the comp range up front.

Lead with their GitHub work, give the comp range, and say plainly that the job is stable. Don't mention Mercor. If they're on a contract, they'll make the comparison themselves.

When should you reach out to engineers on AI-training contracts?

Contact them when a project is ending, not when it starts. Mercor's demand depends on a few buyers, and projects can end overnight, which sends contractors back to the market in groups.

The money is there: Mercor's annualized gross revenue run rate reached $2B in June 2026, up from $1B four months earlier. The risk is where it comes from.

91%

There's already an example. In November 2025, Mercor ended a Meta project called "Musen" that involved 5,000+ contractors. It replaced it with "Nova" at $16/hr, 24% less than the previous $21. A 2026 review reports repeated waves of contributors removed from projects, often with no reason given. Those were generalist rates, not $150/hr senior work, but the same risk applies: if the buyer leaves, the contract ends.

How to time your outreach:

• Don't pitch in the first month of a contract. They just agreed to exclusivity and planned around the income. They will say no.
• Watch for their open-source activity to come back. An exclusive 30 to 40 hour week likely cuts into personal commits. If commits to their own repos pick up again after a quiet stretch, the contract may have ended.
• Follow AI-lab news. With 91% of revenue from AI labs, any lab that cuts eval spending is a reason to contact your warm list that week.
• Keep the list warm. Rerun the same search every month and compare it to last month's results. People who now show a new gap in their history, or who started committing again, go to the top.

FAQ

How much does Mercor pay software engineers?

It depends on the listing. The most senior roles in the news pay $150/hr for open-source maintainers and $120 to $180/hr for senior software engineering experts with FAANG+, Mag 7+ or top quant backgrounds. More typical work pays much less: on Nov 12, Mercor listed backend software engineer work at $50 to $80/hr, and its homepage showed a $90/hr average contracted rate.

Can a Mercor AI trainer software engineer hold a full-time job at the same time?

For the top engineering roles, no. The maintainer listing requires 30 to 40 hours a week, says the role isn't a fit alongside a full-time job, and bans work with any other AI lab or AI training company during the engagement. Lower-paid, part-time projects on the same platforms can be done on the side, but the $150/hr work usually means the person is between employers.

What's the best way to start recruiting open source maintainers?

Start with the repos your product depends on and pull names from their MAINTAINERS and CODEOWNERS files. Then filter for recent activity: 10+ PR reviews or 3+ merged PRs in the last 12 months. Job titles don't work well, because in Refolk's index only 234 profiles across the US, UK, Canada and Germany say "open source maintainer" in the headline.