High-Quality Dataset Creator (Images to Q&A)
Budget: $250 – $750 USD
Hiring: LLM Dataset Creator – Challenging Q&A from Images
We're looking for a sharp, creative thinker to help us build high-quality training data for large language models (LLMs). Your task: given an image from a specific domain (e.g., medical, legal, technical, or abstract visual scenes), craft challenging, nuanced Q&A pairs that test the limits of current AI systems.
What You'll Do:
- Receive a domain-specific image (e.g., a chart, diagram, interface, or real-world scene)
- Create 10 difficult, high-quality questions based on the image
- Provide accurate, well-reasoned answers
- Ensure questions are non-trivial, multi-step, or ambiguous enough to challenge LLMs
- Focus on reasoning, inference, and edge cases—not just surface-level facts
Compensation:
- Fixed payment per accepted Q&A pair
- Paid per question, not per image
- Initial test required (unpaid questionnaire and 3 paid Q&A assessment to assess quality)
Ideal Candidate:
- Strong critical thinking and writing skills
- Background in linguistics, AI, education, or a technical field is a plus
- Comfortable working with abstract or domain-specific visual content
- Understands what makes a question “hard” for an LLM (multi-hop, distractors, subtle logic, etc.)
We're looking for a sharp, creative thinker to help us build high-quality training data for large language models (LLMs). Your task: given an image from a specific domain (e.g., medical, legal, technical, or abstract visual scenes), craft challenging, nuanced Q&A pairs that test the limits of current AI systems.
What You'll Do:
- Receive a domain-specific image (e.g., a chart, diagram, interface, or real-world scene)
- Create 10 difficult, high-quality questions based on the image
- Provide accurate, well-reasoned answers
- Ensure questions are non-trivial, multi-step, or ambiguous enough to challenge LLMs
- Focus on reasoning, inference, and edge cases—not just surface-level facts
Compensation:
- Fixed payment per accepted Q&A pair
- Paid per question, not per image
- Initial test required (unpaid questionnaire and 3 paid Q&A assessment to assess quality)
Ideal Candidate:
- Strong critical thinking and writing skills
- Background in linguistics, AI, education, or a technical field is a plus
- Comfortable working with abstract or domain-specific visual content
- Understands what makes a question “hard” for an LLM (multi-hop, distractors, subtle logic, etc.)