Home/Open roles/Software engineers
56 live listings · updated Sep 25, 2026
Coding is where AI labs spend the most on human review, because model-written code that looks right and is subtly wrong is expensive. Engineers get paid to find exactly that.
The work
Not data entry, and not a job. It is expert review, done through a web dashboard, paid by the hour for the hours you log.
You read a generated patch and judge whether it is correct, whether it handles the edge cases, and whether you would approve the pull request. Then you explain why — the explanation is the training signal.
You author tasks a model should be able to do and often cannot: concurrency bugs, subtle API misuse, performance traps, migrations that need to be reversible.
Given two implementations, you pick the better one and articulate the tradeoff. This sounds easy and is the work most people underestimate — the reasoning has to be legible.
Pay
Every figure here is read from the platform's own listing. Where a listing shows no rate, it is not counted rather than estimated. Based on46 hourly listings open today.
Rates are the starting figure on each listing. The top of a range usually goes to the most senior applicants, and some listings pay per task or per project instead — the board marks which.
Eligibility
Platforms verify credentials, so apply where you genuinely fit. This is who these listings are written for:
Honest caveats
The parts people find out after they have already spent an evening on an assessment:
Getting accepted
The screening is the whole gate. These are the things that decide whether a strong applicant gets matched to a project.
Projects are scoped by language and domain. "Rust, distributed systems, six years" matches; "senior software engineer" does not. List the languages you would defend in review.
This is the field where a degree matters least. Systems you have actually run, incidents you have debugged and code review you have done at scale are the signal.
The screen tests whether you can articulate why one solution is better. Practise saying the tradeoff, not just picking the answer — that skill is the entire job.
Open now
Coding evaluation, code review, and software tasks for AI labs. Refreshed Sep 25, 2026.
FAQ
No. This is the field where demonstrated ability matters most and credentials matter least. Self-taught engineers with strong production experience are regularly accepted; the screen is technical, not a resume check.
Live rates for engineering listings are shown above, from the platform directly. Specialist and senior work sits at the top of the range; general coding evaluation sits in the middle.
Python and JavaScript or TypeScript carry the most volume by a wide margin. Rust, Go, C++ and specialist ML work appear less often but pay more when they do because the reviewer pool is smaller.
No. Data labelling is high-volume categorisation at low rates. This is code review and problem authoring that requires engineering judgement, and it is paid accordingly.
Check your contract for outside-work, moonlighting and IP-assignment clauses. Broad IP assignment terms are the ones to look at carefully. Many engineers do this without issue, but read your agreement.
Yes. Engineering listings have the widest international eligibility of any field on the board, though pay is sometimes tiered by region on the lower-rate platforms.
Comparable to a mid-to-senior industry interview, with more weight on explaining tradeoffs than on producing an optimal solution quickly. Several platforms allow only one attempt, so take it when you are sharp.
For most people, no. Weekly hours are usually capped and project availability is uneven. Treat it as strong side income rather than a salary replacement.
Get new engineering roles by email
One short email when something opens in engineering. There are 56 open right now. No digest spam, unsubscribe in one click.