LLM Model Response Evaluation

Lifted (an Upwork Company); United States; À distance; USD 70 000 - 100 000; Plein temps

Avantages offerts par ce poste

Flexible schedule

Remote work

Résumé du poste

Lifted, an Upwork Company, is seeking evaluators skilled in LLM and GenAI data assessment to help verify model outputs across multiple modalities. You will follow structured rubrics and research unfamiliar topics using trusted sources to ensure quality and reliability.

This remote role offers flexibility and a variable workload, with tasks available across arts, culture, history, science, and more. A Bachelor's degree is required, and prior experience with multimodal evaluation is preferred.

Qualifications

• 3+ years of hands-on LLM/GenAI data evaluation.
• Bachelor's Degree required.
• Ability to research unfamiliar topics using trusted sources.

Responsabilités

• Evaluate model-generated content against quality rubrics (factuality, consistency, aesthetics).
• Assess content across text, images, audio, video, HTML widgets, PDFs.
• Follow detailed project guidelines and document findings.

Connaissances

LLM data evaluation

GenAI evaluation

Research skills

Multimodal evaluation

Formation

Bachelor's Degree

Description du poste

• Evaluating UI widgets, infographics, image factuality, side-by-side comparisons, and similar AI evaluation activities.
• The work may involve text, images, audio, video, HTML widgets, PDFs, or combinations of these modalities.
• The work is domain-agnostic and may cover topics across arts, culture, history, science, engineering, and more.
• Resources will be expected to independently research unfamiliar topics using trusted sources before making evaluation decisions.
• Each task will include detailed project guidelines within the evaluation platform.
• 3+ years of hands-on experience in LLM / GenAI data evaluation.
• Bachelor's Degree required.
• Ability to research unfamiliar topics using trusted sources and make well-supported judgments.
• Comfortable evaluating content across multiple modalities.
• Flexible and remote work.
• Variable workload: Accept or decline tasks based on your availability.
• No guaranteed hours: Workload may vary weekly.

Our client, a global technology company that helps businesses build, train, and manage AI systems is looking for experts to evaluate model-generated content against defined quality rubrics such as factuality, consistency, aesthetics, and other evaluation criteria.