The work
You will evaluate AI chat and search tools under consistent conditions and judge whether their responses are accurate, helpful, and safe. You will use detailed guidelines to rank and annotate outputs, then provide structured feedback.
The role also includes testing prompts across different user needs and edge cases, checking whether evaluations match the guidelines, recording findings, and submitting evaluation datasets. You will work with other trainers and contributors to improve instructions and keep ratings consistent.
- Evaluate AI chat and search responses using detailed grading rubrics.
- Rank, annotate, and provide feedback on model outputs.
- Write, refine, and stress-test prompts for varied scenarios and edge cases.
- Identify rating and rubric discrepancies and correct them.
- Record findings and submit evaluation datasets using AI training tools.
- Take part in quality reviews and help maintain evaluation consistency.
What it pays and takes
This is a remote, part-time contractor role for candidates in Germany. Experience with AI training or evaluation work is preferred, but strong judgment, careful review skills, and clear communication are central to the work.
- Pay: $14 to $36 per hour.
- Location: Open to candidates in Germany.
- Language: Fluent in English.
- Level: Intermediate.
- Work type: Part-time contract work.
- Strong attention to detail when reviewing AI-generated content.
- Clear written and verbal communication for reporting and collaboration.
- Analytical problem-solving and sound judgment when evaluating model behavior.
- Good time management and self-organization for independent remote work.
- Preferred: Experience with AI model training, RLHF, data labeling, prompt engineering, evaluation rubrics, or related projects.
- Helpful: Experience with machine learning data preparation, large language model projects, or data-focused AI evaluation.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind many AI systems. People review examples, rate model responses, test prompts, and provide feedback so models can produce more useful and reliable results.