About the job

Overview: We are looking for evaluators from Bangladesh proficient in Urdu language to review and compare AI-generated responses. The role focuses on identifying toxic or harmful content across native scripts and transliterated text and assessing model performance across multiple datasets.

Rate per hour: 1095 PKR per hour
Minimum Hour Commitment per day: 5-6 hours per day
Work Mode: Freelancing (Remote)

Key Responsibilities:

Evaluate AI model outputs in Urdu.
Identify and flag toxic, harmful or hate-based content, including subtle or context-dependent cases
Compare model responses and provide performance assessments based on predefined criteria
Classify the type and severity of toxicity, e.g. hate speech, harassment, abusive language
Provide brief explanations for flagged items where required
Ensure consistency, accuracy and adherence to project guidelines

Qualifications/ Required Skills:

Proficient in English and Urdu.
Minimum 1 year of experience in content writing, content moderation, linguistic evaluation or a similar domain
Strong understanding of cultural nuances, slang and context-dependent expressions
Ability to identify toxicity in both native script and transliterated formats.
Good analytical and evaluative skills
Prior experience in content moderation, linguistic evaluation or data annotation is a plus.
Based in Pakistan.

Education

Minimum of a bachelor’s degree in any field, i.e. Humanities, Linguistics, Mass Communication or related disciplines.

To apply for this job please visit www.linkedin.com.