Software Engineer (AI Training)

Write and assess software for AI training

Build solutions and evaluate generated code to improve how AI coding assistants reason about software. This opportunity is directed toward software engineers in Shanghai and combines hands-on programming with detailed technical review.

• Write, debug and optimize solutions across programming languages, frameworks and problem types.
• Assess generated code for correctness, efficiency, readability and sound engineering practices.
• Identify logic defects, security problems, bugs and performance constraints in model responses.
• Explain why a solution succeeds or fails through detailed, organized feedback.
• Create difficult coding prompts and edge cases that test the limits of AI capabilities.
• Compare alternative generated solutions and rank them by engineering quality.
• Manage assignments independently on an asynchronous schedule.

Core engineering skills

Proficiency in at least one programming language is required, such as Python, JavaScript, Java, C++, Go or TypeScript. Candidates need a solid foundation in data structures, algorithms and software design principles.

You should be able to read, write and reason through code precisely, with strong problem-solving skills and a methodical approach to debugging. Clear written explanations are important: technical decisions need to be understandable without unnecessary complexity.

A bachelor's degree in Computer Science, Software Engineering or a related subject is required, or equivalent practical experience. Reliable independent work is essential to this contract arrangement.

Helpful additional experience

Broader experience across full-stack, backend, systems or data engineering work is valuable. Familiarity with AI/ML, language models or prompt engineering is also useful, as is previous code review, technical writing or mentoring of junior developers.

Other strengths include cloud services, DevOps, distributed systems, competitive programming, open-source contributions and personal projects that demonstrate engineering ability. Graduate study in Computer Science or a related discipline is an advantage.

Working arrangement

The fully remote hourly contract calls for 10–40 hours per week. Scheduling is flexible and asynchronous, allowing you to organize your own work while contributing to AI research projects.

The work pairs freelance independence with varied technical problems and direct assessment of AI-generated code. Further assignments or contract extensions may become available when new projects launch.