Project Overview
We are sourcing independent Scientific and Technical Service Specialists to provide their expertise for an AI benchmark evaluation project. As AI models increasingly generate professional-grade scientific analyses, technical reports, and STEM-focused deliverables, their accuracy relies entirely on robust, expert-crafted training data. The objective of this project is to autonomously produce high-quality evaluation tasks, strong prompts, and clear, well-structured rubrics that generate clean, reliable data for model training.
Project Deliverables & Scope
Operate autonomously to design complex evaluation frameworks and provide structured training data. Expected deliverables include:
- Task & Prompt Creation: Generating realistic, high-quality prompts that compel the AI model to produce complex, professional-grade deliverables specific to scientific research, engineering, and technical services.
- Rubric Development: Writing clear, well-structured evaluation rubrics with criteria that are highly specific, non-ambiguous, and easy to score.
- Benchmark Evaluation Data Generation: Producing clean, reliable training data that directly aids in the evaluation and refinement of AI models handling complex STEM-related tasks.
- Quality Assurance & Fact-Checking: Ensuring all generated tasks and scoring criteria reflect strict, real-world scientific methodologies, technical standards, and empirical accuracy.
Required Expertise
To successfully fulfill the deliverables of this project, Contractors must possess deep industry knowledge to craft realistic professional scenarios.
Core skillset includes
- Demonstrable professional expertise within scientific, engineering, or technical service sectors, with a deep understanding of industry standards, technical terminology, and high-level analytical deliverables.
- Strong writing and prompt generation skills, with the ability to design highly realistic, complex STEM task scenarios for AI evaluation.
- Proficiency in rubric generation, specifically the ability to create objective, non-ambiguous scoring criteria that leave no room for subjective interpretation.
- A meticulous, detail-oriented approach to fact-checking advanced scientific literature, data interpretation, and technical specifications to generate reliable data for system benchmarking.
We offer a pay range of $10-to-$30 per hour, with the exact rate determined after evaluating your experience, expertise, and geographic location. Final offer amounts may vary from the pay range listed above. As a contractor you’ll supply a secure computer and high‑speed internet; company‑sponsored benefits such as health insurance and PTO do not apply.
Engagement Type: Freelance / Independent Contractor
Workplace Type: Remote