Software Engineer, AI Coding Task Creator

You will create challenging coding tasks that evaluate whether AI models can handle real software engineering work. Tasks may involve bugs, feature requests, refactoring, performance problems, or other challenges drawn from repositories you know well.

Each task must run in a reproducible environment and include reliable checks for correctness. You will also provide a reference solution and identify tasks that are too unclear or difficult to verify consistently.

• Create realistic coding problems based on familiar repositories and real engineering challenges.
• Design tasks that require judgment rather than lookup work or boilerplate code.
• Build reproducible development environments for each task.
• Write verifiers that accept different correct implementations and reject incomplete, incorrect, or hard-coded solutions.
• Provide a valid reference solution and flag ambiguous tasks.

Previous AI work is not required, but interest in AI and its technical challenges is useful.

• Time: Less than 20 hours per week.
• Languages: Professional fluency in English.
• Technical skills: Hands-on expertise in at least one of Python, Java, Rust, Go, C++, or TypeScript.
• Knowledge: Strong understanding of algorithms, data structures, and software engineering principles.
• Experience: Ability to debug complex systems, fix bugs, deliver effective optimizations, and explain technical decisions clearly in documentation.