GitHub Contributor

What you'll do

You build reinforcement learning environments, reproducible setups that test whether an AI model can solve a real software engineering problem. Each one pairs a task drawn from real code (a defect to fix, a feature to implement, a refactor, a performance bottleneck) with a golden reference solution the model gets graded against.

Alongside that you contribute expert-level code samples and debugging strategies, document the reasoning behind your solution, and review other contributors' submissions for accuracy.

The engagement is remote contractor work at roughly 15 hours a week. No prior AI experience is required; the domain knowledge is what is being paid for.

What you need

• Clear open-source contributions on a public profile, GitHub or GitLab. This is the one hard requirement, and C++, Python, Java, Go, TypeScript and Rust are named as the preferred languages to show them in
• Availability to start your first tasks within 24–48 hours of finishing onboarding
• Around 15 hours a week

Everything in the next section is filed under "preferred" rather than "required", unusually loose phrasing for a senior contract. Read it as the bar the strongest applicants will clear.

Nice to have

• Significant hands-on expertise in at least one of Python 3, Java, Rust, Go, C++ or TypeScript
• Deep understanding of algorithms, data structures and software engineering principles
• A track record of debugging complex systems and delivering real optimisations
• Experience refactoring large codebases and modernising legacy systems
• High-impact features delivered end to end on cross-functional projects
• Strong technical documentation and communication
• Curiosity about AI; previous AI experience is stated as not required

What it pays

$50–150/hour, but read the compensation section before planning around that. The engagement is output-based: experts are paid per task that meets the project specification, not per hour worked. The hourly band is what the board displays; what reaches your account is task count times task rate, so your effective rate depends on how fast you can produce an acceptable environment.

A minimum weekly submission requirement also applies, and the number is missing from the ad: it reads "Experts must submit a minimum of tasks per week". Ask what it is before you accept, because it is the difference between a flexible side engagement and a quota.

How you get hired

Four stages: apply and answer the screening questions, complete an AI interview of about 30 minutes, sit a technical assessment (marked tentative), and pass a hiring manager review. None of that time is paid.

Roles are stated to fill typically within 48 hours, with a selected expert starting tasks 24 to 48 hours after onboarding, so this is not a listing to apply to and forget about.

Worth knowing

Good:

• $50–150/hour sits at the top of this market, and the band is published up front
• Your existing open-source record is the credential; no AI experience and no specific degree are asked for
• The work is genuine engineering: reproducible environments and reference solutions, not labelling
• 100 openings, so this is a volume hire rather than a single seat
• Roughly 15 hours a week, shaped to sit alongside other work

Less good:

• The advertised hourly band and the actual output-based pay structure are two different things, stated in the same ad without being reconciled
• A weekly submission minimum applies, but its number is missing from the listing
• A $100-wide band with no published basis for where in it you land
• Four screening stages, including an unpaid technical assessment, before any paid work
• Contractor engagement: no benefits, no notice period, your own tax to manage
• Starting 24–48 hours after onboarding is hard to combine with a notice period elsewhere

About this listing

Posted by micro1 on its own jobs portal, confirmed open on 7 September 2026. The $50–150/hour band, the process, and every requirement above are its own published text for this listing.

No country restriction is named on the ad itself, so eligibility is left open. We could not verify the per-task rate or the weekly submission minimum; neither is published. Screening and payment run at the platform level. See micro1.