Remote Life Sciences AI Evaluation Designer

Gramian Consulting; United States; Remote; PKR 1,378,000 - 2,755,000; Part time

Job summary

Gramian Consultancy is seeking experienced Life Sciences researchers to design advanced scientific reasoning tasks for an AI evaluation benchmark focused on biology and related disciplines.

The role is a non-coding research position focusing on creating self-contained problems from realistic experiments, datasets, figures, and literature-style evidence to test interpretation and synthesis of scientific information.

Qualifications

• PhD or equivalent advanced research in Life Sciences.
• Deep expertise in at least one biological discipline.
• Strong ability to interpret scientific experiments, data and findings.
• Strong understanding of experimental design and scientific reasoning.

Responsibilities

• Design challenging life-sciences reasoning problems for AI evaluation.
• Create tasks based on realistic experiments, biological datasets, figures, sequences, structures and scientific evidence.
• Write questions with clear, objectively defensible answers.
• Develop rigorous expert reference answers explaining the required scientific reasoning.
• Ensure tasks test interpretation and synthesis, not simple factual recall.
• Validate biological claims, calculations, assumptions and conclusions.
• Identify ambiguity, alternative valid interpretations and poorly constrained questions.
• Build tasks that differentiate advanced AI models based on scientific reasoning ability.
• Incorporate reviewer feedback while maintaining high quality and throughput.

Skills

Scientific reasoning

Literature analysis

Education

PhD or equivalent advanced research in Life Sciences

Job description

Gramian Consultancy is seeking experienced Life Sciences researchers to design advanced scientific reasoning tasks for an AI evaluation benchmark focused on biology and related disciplines.

The role is a non-coding research position focusing on creating self-contained problems from realistic experiments, datasets, figures, and literature-style evidence to test interpretation and synthesis of scientific information.