Research Engineer (Code Generation and Model Evaluation)

You will help improve AI systems by writing and reviewing software, creating coding tasks, and evaluating model outputs. The work uses real-world code and requires careful technical judgment, clear feedback, and well-documented decisions.

• Analyze, debug, and resolve issues in Python, Java, Rust, C++, Go, and TypeScript codebases.
• Build features for code-generation workflows and refactor software for performance, maintainability, and adaptability.
• Create and review coding tasks used to measure AI model performance and accuracy.
• Write clear annotations, technical explanations, and supporting documentation.
• Validate code carefully and provide actionable feedback on model-generated solutions.
• Work with technical stakeholders in a remote, collaborative setting.

What it pays and takes

The work is output-based, and minimum weekly submission requirements apply.

• $50 to $100 per hour equivalent, paid per task that meets project specifications.
• Less than 20 hours per week.
• Location: Open to candidates in Algeria, Argentina, Austria, Bahrain, Bangladesh, Belgium, Brazil, Canada, Chile, Colombia, Denmark, Egypt, France, Germany, Ghana, Greece, India, Indonesia, Italy, Japan, Jordan, Kazakhstan, Kenya, Kuwait, Kyrgyzstan, Lebanon, Malaysia, Mexico, Morocco, the Netherlands, Nigeria, Oman, Pakistan, Palestinian Territories, Peru, the Philippines, Portugal, Qatar, Saudi Arabia, Singapore, South Africa, Spain, Sri Lanka, Switzerland, Taiwan, Thailand, Tunisia, the United Arab Emirates, the United Kingdom, the United States, Uzbekistan, and Vietnam.
• Fluent in English.
• Strong knowledge of algorithms, data structures, competitive programming, and coding problem analysis.
• Highly capable in at least one of Python, Java, Rust, C++, Go, or TypeScript.
• Strong analytical skills, attention to detail, careful code validation, and clear written and verbal communication.
• Ability to work independently, collaborate remotely, and deliver high-quality, well-documented code.
• Experience as a problem setter, tester, or checker on a major competitive programming platform is helpful.
• Open-source contributions or experience with collaborative software projects is helpful.

How it works

You apply here in about two minutes and finish on the hiring platform. The next step is a short AI interview (technical roles include a coding challenge). Passing it is how you qualify for paid tasks, so take it soon after you are invited.

About AI training work

AI training is the human work behind systems that generate and understand code, text, images, and other content. Engineers and other specialists review examples, test model outputs, and explain what good results should look like so models can improve.