Open Source Won't Pay You. This Coding Side Hustle Pays $150/hr

AI labs are paying senior software engineers $50 to $150 an hour to work on open source repositories. Right now, two separate micro1 listings have 100 openings each for engineers with public GitHub histories. If you want a viable coding side hustle, this changes where your hours should go.

Most developer advice pushes you to build a micro-SaaS, record a video course, or hunt bug bounties. Those ideas force you to work for weeks before anyone decides whether to pay you.

This guide sorts every coding side hustle by who decides whether you get paid. You will see what beats your day-job rate, why automated coding agents altered traditional issue bounties, and how to convert public pull requests into reliable hourly income.

Disclosure: Skillora earns a referral fee from micro1, Mercor and Alignerr if you sign up through our links and work on their platforms. It costs you nothing and does not change the rates quoted, which come from the platforms' own listings.

Get paid to contribute to open source: the two listings worth reading first

micro1 has opened two separate contractor listings that hire engineers specifically to work on open source codebases. Each role has 100 openings.

The first role is GitHub Contributor. It pays $50 to $100 per hour. The position requires roughly 15 hours per week of remote contractor work. You build reinforcement learning environments that test an AI model's ability to solve real software engineering problems. These problems include fixing bugs, implementing features, refactoring existing architectures, and optimizing runtime performance. For every problem, you build a reproducible test environment and supply a "golden reference solution". You also review and validate code written by other domain experts. The listing covers Python, Java, Rust, Go, C++, and TypeScript. According to micro1, applicants "must have clear open source contributions and profiles to showcase it like GitHub or GitLab". micro1 states that no prior AI experience is required. The hiring pipeline involves an initial application with screening questions, an AI interview that lasts about 30 minutes, a tentative technical assessment, and a hiring manager review.

The second role is Open Source Contributor (GitHub). It pays $100 to $150 per hour. This role also has 100 openings, runs globally as a fully remote contract, and asks for about 15 hours per week on a flexible schedule that includes weekends. In this job, you explore unfamiliar repositories, implement and debug front-end and back-end components, reproduce issues, and isolate root causes. You write unit, integration, regression, and end-to-end tests. You inspect AI-generated code patches for correctness, completeness, security, and maintainability. You also translate full software builds from one programming language into another. To qualify, you need a public or otherwise verifiable GitHub profile showing meaningful ownership of or contributions to open source repositories, along with multi-language experience. micro1 emphasizes that contributions must demonstrate genuine engineering competence, "not only documentation updates, dependency bumps, formatting changes, or minor typo corrections". Candidates must prepare to discuss specific repositories, pull requests, issues, and code they personally authored. The evaluation process includes the application, a 30-minute AI interview, and a hiring manager review.

Across those two listings the band is $50 to $150 an hour, with 100 openings each.

At the top end of the market sits the Open Source GitHub Maintainer role. This role pays $150 to $300 per hour across 10 openings. It requires about 15 hours per week and accepts applicants globally. Your job is to review patches that a large language model generates for the exact repositories you maintain. You deliver professional assessments with written justifications for every patch, and peer review work submitted by other subject-matter experts. The AI interview for this role takes about 10 minutes.

Now consider the catch. Across all three micro1 roles, compensation is output-based. Experts are paid per task that meets the project specification. The time a task takes varies, and minimum weekly submission requirements apply. So the hourly figure is a rate, not a guarantee of hours. micro1 also notes that it typically fills active roles within 48 hours and expects newly onboarded experts to begin working on tasks within 24 to 48 hours.

Why AI labs pay for your GitHub history in 2026

AI models need human-verified engineering judgment, and a years-long public record of merged human work is the proof of it. AI agents now join public contribution queues, but they cannot produce a multi-year human track record. That human record is what micro1 buys, which is why typo fixes and dependency bumps do not qualify.

At the same time, maintainers are closing doors to open contributions. In January 2026, tldraw began automatically closing pull requests from external contributors, citing a rise in low-quality AI-generated pull requests. That same month, Daniel Stenberg ended the bug bounty program for curl to remove the financial incentive for low-quality reports, noting that many were generated by AI tools.

Public bug programs are pulling back as well. On 18 May 2026, HackerOne cut reward payouts across its Internet Bug Bounty program and paused new submissions.

Meanwhile, traditional open source rarely pays maintainers directly. According to Tidelift's 2024 State of the Open Source Maintainer report, 60% of open source maintainers work as unpaid hobbyists, and only 12% earn most or all of their income from maintaining.

AI training marketplaces pay $50 to $300 an hour for that exact engineering history. They need developers who understand architecture, edge cases, and regression risks to evaluate models against real code.

The bar every software engineer side hustle has to clear

Before taking on extra work, look at your primary compensation. The Bureau of Labor Statistics reported in May 2025 that median software developer pay reached $135,980 per year. Based on a standard 2,080-hour work year, that salary equals roughly $65 an hour.

If you work as a senior or staff software engineer, your standard rate sits higher. Taking on moonlighting work that pays less than $65 an hour pays less than your day job. Low-paying work consumes mental energy while producing less money than your primary employer pays for the same block of time. Every software engineer side hustle must clear that $65-an-hour floor after factoring in platform fees and unpaid admin time. If a side job cannot beat $65 an hour, skip it.

Coding side hustles sorted by who decides you get paid

Side hustles do not fail on technical execution. They fail on payment mechanics. Most lists rank ideas by technical category. A better approach groups every side job by the authority that controls your payout:

1. Paid for accepted work. A client sets a spec. Work that meets the spec gets paid.
2. Paid if you win. Many people do the work, one gets paid.
3. Paid if strangers buy. You build first and the market decides later.

Understanding which tier you enter tells you how much unpaid risk you carry before writing any code.

Tier one: paid for accepted work

Tier one offers direct risk alignment. You trade engineering hours for contracted compensation. If your work meets the stated spec, payment follows.

AI model training and code evaluation

AI data platforms hire senior engineers to build environments, draft reference code, and review model outputs. There is no bidding. You pass one screen, then tasks arrive with the rate attached, though pay is per accepted task:

• micro1 software roles: Typical listings run $50 to $100 per hour, with the Maintainer listing reaching $300 per hour. A specialized Senior Software Engineer role pays $245 to $280 per hour. Review our micro1 review for an in-depth breakdown of the onboarding flow.
• Mercor software listings: Typical pay sits at $70 to $98 per hour, with top-tier roles reaching $210 per hour. Its Frontend, Backend, Full-Stack, and DevOps/Platform Engineer talent networks each list $70 to $150 per hour. A dedicated Legacy Codebase Migration Expert role lists at $200 per hour. Read our Mercor review for screening details.
• Alignerr software engineering roles: Typical positions pay $50 to $90 per hour, with top listings reaching $120 per hour. See our Alignerr review to evaluate their assessment tests.
• Complete software engineering directory: Track open listings across every platform using Skillora's curated engineering directory, or review the best AI training jobs guide.

AI training is the only option here that clears the $65-an-hour baseline without unpaid bidding.

Contract freelancing

Traditional freelancing also lives in tier one, but market friction is higher. According to Upwork's official hiring guide, software developers on Upwork cost $10 to $100 an hour across three tiers:

• Entry level: $20 to $40 per hour
• Intermediate: $40 to $70 per hour
• Expert: $70 to $150+ per hour

Since 1 May 2025 Upwork charges freelancers a variable service fee of 0% to 15% per contract, fixed for the life of the contract.

Only Upwork's expert tier clears the $65 bar. You must also win the contract against global bids, which pushes freelancing partly into tier two. Drafting proposals, reviewing client postings, and participating in introductory calls is uncompensated labor. If you spend five unpaid hours bidding to secure a single three-hour gig, your real hourly earnings drop.

Tier two: paid if you win

Tier two side jobs turn development into an unpaid contest. You complete technical work upfront, but you only earn money if the sponsor picks your submission over every other entrant.

Bug bounty platforms

Security research sounds lucrative, but money concentrates heavily in the top programs. HackerOne reported that it paid $81 million in bug bounties in the 12 months to 30 June 2025, up 13%. The top 100 programs paid $51 million of it.

Rewards for open source bugs have dropped sharply. On 18 May 2026, HackerOne cut its Internet Bug Bounty rewards: critical $9,250 to $2,257, high $4,429 to $1,009, medium $1,843 to $297, and low $597 to $68. The program was also paused and not accepting new submissions.

Meanwhile, curl ended its bug bounty at the end of January 2026 to remove the financial incentive for low-quality reports. The race now includes AI agents, and the organisers are cutting prizes.

Open source issue bounties

Platforms like Algora pay cash rewards when your pull request gets merged. A repository maintainer posts an issue with a bounty attached. You clone the repo, solve the issue, and open a PR.

The structural problem is competition. Many contributors, and now AI agents, race for the same issue. The maintainer merges only one pull request. Only the merged PR gets paid. If the maintainer merges someone else's fix, your time earns nothing.

Tier three: paid if strangers buy

Tier three carries real upside but zero income floor. You build systems with no guarantee that anyone will ever buy what you produce. The market decides your payment long after your development work is done.

SaaS and apps

Building software products offers uncapped upside, but it pays nothing until it sells. You work for months before the market votes. A solo developer can build an entire product, configure payments, and provision servers without securing a single paying subscriber. Technical execution does not create customer demand. Tier three products demand ongoing marketing, support, and sales work. It works as a project, but it is poor as reliable side income.

Technical video courses

Selling video courses seems passive, but marketplace platforms control distribution and take substantial cuts. Udemy's November 2023 instructor letter filed with the SEC details the platform's revenue distribution. Udemy pays instructors 37% of organic marketplace sales.

Its subscription revenue share has decreased steadily:

• January 2024: cut to 20%
• January 2025: cut to 17.5%
• January 2026: cut to 15%

You absorb the upfront production cost while the instructor's share of subscription revenue shrinks.

GitHub Sponsors

Relying on donations or community sponsorships for open source rarely pays commercial rates. Tidelift confirmed that 60% of open source maintainers remain unpaid hobbyists, while only 12% earn most or all of their income from maintaining. Without corporate backers, GitHub Sponsors functions as an occasional tip jar rather than regular compensation.

Side jobs for developers: which one fits your GitHub

To find viable side jobs for developers, match your public profile to the platform that values it most:

What your employer can claim

Before starting moonlighting work, check your legal boundaries. Your employment agreement may claim what you build (invention assignment) and may restrict paid outside work (moonlighting clauses). For a detailed breakdown of legal boundaries, read our guide on intellectual property and employment agreements. This summary provides general information, not legal advice.

AI training platforms limit intellectual property risk because you work within tasks rather than building a standalone product. You build nothing an employer can claim. However, confidentiality rules still apply. Never paste proprietary employer code, internal docs, or customer data into a task. Write entirely from your own knowledge and public codebases.

If your employer builds foundation models or sells to AI labs, read your outside-activity policy and obtain written clearance before accepting work.

For open source listings specifically, the GitHub history you show micro1 must be your own public work. Do not present code you wrote on the clock for an employer as personal contribution unless it was released publicly under the employer's open source program.

How to pass the micro1 screen

micro1 screens technical candidates using an automated AI interviewer. Contributor listings require an interview lasting about 30 minutes. The Open Source Maintainer listing uses a 10-minute screening conversation.

The listing tells you to be ready to discuss specific repositories, pull requests, issues, and code you personally authored.

Prepare for the AI interview with these concrete steps:

1. Select your three strongest merged pull requests from your public GitHub history. Skip simple formatting changes, typo fixes, and dependency bumps.
2. Review the core problem for each pull request. Be ready to explain the original bug or feature clearly.
3. Outline the alternatives you rejected. Be prepared to explain why your approach solved the problem better than other options.
4. Describe the reviewer pushback you encountered. Explain what the reviewer asked you to change and how you responded.

To prepare for this automated screening format, study our micro1 AI interview guide. Review how evaluation pipelines operate across providers in our guide to AI training platform screening order. To practice answering technical questions under automated conditions, use the Skillora mock interview tool.

The short version

Traditional side hustles push you into unpaid speculative work. Bug bounties are shrinking, freelance bidding wastes unbillable hours, and products offer zero income floor. AI labs now provide an alternative by paying senior engineers $50 to $150 an hour for verified open source work, with 100 openings each on two micro1 listings. Make sure your moonlighting clears your $65-an-hour salary baseline, protect employer confidentiality, and pick roles where you are paid for accepted work rather than speculative bids. Explore open roles today on our software engineering AI training job board.