Get paid to put your 
coding expertise to work on frontier AI
Apply
Get verified
Match with tasks
Start earning
How your coding expertise shapes more reliable AI
Frontier AI still struggles with problems that require deep engineering reasoning. Your expertise helps AI teams identify where models fail, create better training data, and measure whether new models are actually improving. Work alongside other experts on projects where rigor, accuracy, and sound judgment matter.
Create
Review
Evaluate & improve
Getting started
Apply
Tell us your background and expertise.
Get verified
We confirm your identity and credentials via a background check. This is what makes companies trust every expert in our community.
Get matched
Get paid
What kinds of coding expertise are we looking for?
Python & Linux/Unix Systems
Bring deep fluency in Python, shell scripting, Linux/Unix environments, containerization (Docker/Kubernetes), and package/environment management. You may spin up a dev environment from scratch, diagnose a broken build, or automate the fix.
Backend & Systems Engineering
Draw on experience with server setup, system administration, data pipelines, and infrastructure. You may chain together multiple tools, services, and configurations to produce a working, reproducible result end to end.
DevOps, SRE & Platform Engineering
Apply a background in DevOps, site reliability engineering, or platform engineering. You may design tasks that blend automation, CI/CD pipelines, monitoring, and infrastructure-as-code with rigorous, verifiable outcomes.
Security & Code Quality
Use experience in security/penetration testing, code review, or evaluating competing implementation approaches. You may spot a suboptimal solution, suggest improvements, or judge whether an alternative approach is more robust.
What the work actually looks like
Complete a complex, real-world, end-to-end challenge in a terminal environment.
Provide a clear, reproducible solution with automated verification criteria.
Design it so an AI coding agent actually has to reason through the problem, not pattern-match.
Security & code quality
Other tasks ask you to review a peer's solution or an AI agent's approach, catching subtle errors in logic, configuration, or architecture and judging whether an alternative path is more correct or more efficient.
Expert-level by design
Every task is calibrated so that an expert-level engineer can solve it confidently, but so that it genuinely challenges — and often stumps — today's most advanced AI coding agents.
Engineering judgment over shortcuts
That means multi-step, end-to-end execution over isolated scripting challenges, production-grade rigor over quick hacks, and real engineering judgment over textbook examples. If you're the kind of software engineer who gets bored by easy problems, this is closer to the ceiling of what a task can ask of you.
Why Snorkel AI
Do work that actually matters
Your engineering judgment directly shapes how frontier AI agents handle complex, real-world terminal tasks — not busywork that gets thrown away.
No AI experience required
What matters is deep expertise in software engineering. We'll never ask you to learn a new field to qualify.
Get paid for your expertise, not your hours
Pay is task-based, with rates varying by task complexity and experience.
Total flexibility
This is remote and flexible work — hours depend on the assignment, not a fixed weekly minimum.
Built for real expertise
We look for engineers with real, hands-on software engineering experience — professional experience matters more than a specific degree. Open to eligible U.S.-based contractors; residents of California are not currently eligible.
A vetted community, and your choice, always
We verify every expert's identity and credentials for real, and participation in any project is entirely voluntary.
