The work
You will help evaluate and improve Bengali content produced by AI models. The work combines language review, writing, fact-checking, and careful judgment about meaning, tone, safety, and cultural fit.
- Review Bengali AI content for accuracy, fluency, grammar, spelling, tone, cultural fit, and prompt adherence.
- Find mistranslations, meaning errors, ambiguity, register problems, logical inconsistencies, and misleading claims.
- Fact-check claims when needed and write clearer rewrites or high-quality model answers.
- Create prompts and reference Bengali answers for general knowledge, customer support, education, and conversational writing.
- Compare and rank AI outputs for correctness, fluency, tone, cultural appropriateness, safety, and reliability.
What it pays and takes
This is remote contractor work for multilingual AI training and model evaluation. The project is expected to last about one to three months.
- Pay: $15 per hour.
- Time: Approximately 20 hours per week.
- Location: Open to candidates in Bangladesh, India, Saudi Arabia, or the United Arab Emirates.
- Language: Excellent Bengali writing and editing ability, plus English reading ability for mixed-language work.
- Experience level: Intermediate.
- Required background: Experience with translation quality review, localization, copyediting, proofreading, or grammar and style checking.
- You should be able to judge tone and register for different audiences, identify hallucinations or misleading content, and review factual claims.
- Helpful background includes AI training, model evaluation, language-quality work, writing, translation, editing, linguistics, or familiarity with Bengali regional variation.