The work
You will evaluate bilingual AI-generated content and create training examples that show accurate, clear, and natural responses. The work covers Hebrew language quality, reasoning, factual accuracy, and adherence to instructions.
- Review Hebrew and English AI responses for accuracy, clarity, fluency, and prompt adherence.
- Evaluate step-by-step reasoning and identify methodological, conceptual, or explanatory gaps.
- Fact-check claims with reputable sources when task guidelines require it.
- Write prompts, model solutions, and explanations that demonstrate correct methods.
- Rate and compare responses for correctness, reasoning quality, relevance, and language quality.
- Test for inaccuracies, bias, meaning drift, and other reliability issues.
What it pays and takes
- Pay: $50 per hour.
- Hours: 20 or more per week, with flexible scheduling.
- Location: Fully remote and open to candidates in Israel.
- Languages: Fluent Hebrew with strong formal and informal writing and editing skills; at least C1 English proficiency.
- Education: Bachelor's degree or higher in linguistics, translation, Hebrew language, communications, journalism, or a related field.
- Experience: At least three years in translation, localization, editorial quality assurance, content quality, or linguistic review.
- You must apply detailed evaluation rubrics consistently and identify reasoning gaps and methodological errors even when the language is fluent.
- Helpful background includes AI training, annotation, or evaluation workflows, plus familiarity with Israeli cultural context and terminology in news, education, consumer, and technology topics.
- Work arrangement: Part-time contractor work with independent, asynchronous collaboration.