The work
You will evaluate AI-generated Estonian text and create language data that helps improve AI systems. Your work will cover general knowledge, education, customer support, translation, localization, conversational writing, and culturally sensitive content.
You will use detailed language judgment to rate responses, compare alternatives, write reference answers, and explain errors clearly. Some tasks involve fact-checking claims and identifying safety or policy risks.
- Review grammar, spelling, fluency, meaning, terminology, tone, register, and cultural fit.
- Find mistranslations, ambiguity, unnatural phrasing, incorrect case forms, factual hallucinations, bias, unsafe content, and policy-risk language.
- Write prompts, high-quality Estonian answers, rewrites, and example responses.
- Rate and compare AI responses for correctness, meaning preservation, structure, safety, tone, and overall quality.
- Provide precise written feedback and recommended improvements.
What it pays and takes
This is a part-time contract role for language-focused AI evaluation. It is open to candidates in Estonia.
- $35 per hour.
- 20+ hours per week.
- Fluent Estonian with strong reading and writing ability.
- Professional English proficiency at C1 or higher for following guidelines and writing detailed feedback.
- Deep knowledge of Estonian grammar, spelling, punctuation, cases, word order, compound words, verb forms, tone, and style.
- Ability to distinguish formal, neutral, conversational, professional, educational, and public-facing language.
- Strong attention to subtle meaning changes, literal translations, terminology problems, culturally mismatched wording, and language interference.
- At least three years of experience in writing, editing, localization, translation, content quality assurance, journalism, education, or another language-focused field.
- A degree in linguistics, Estonian language or literature, English literature, communications, journalism, translation, or a related field is helpful.
- Experience with AI data training, translation quality assurance, editorial quality assurance, localization quality assurance, or rubric-based evaluation is preferred.
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 systems that generate and understand language. Evaluators review examples, write better responses, and give structured feedback so models can produce more accurate, natural, and useful results.