Prompt Writer & Rubric Designer | Project Pavilion (Contractor)
Text
Authored deep-research prompts and designed multifaceted scoring rubrics for evaluating AI responses to consumer product queries. Developed factual-error detection, anti-hack criteria, and ensured tight rubric alignment with grounding URLs across multiple domains. Integrated multimodal tasks requiring image identification before text-based product analysis.
• Created rubrics with 8–20 weighted criteria including grounding URLs and criterion tags.
• Designed and validated fact-aligned, anti-hallucination, and negative scoring rubric components.
• Developed self-contained, timeless evaluation criteria and prompt–rubric alignment checks.
• Crafted prompts and tasks for both text- and image-based AI analysis in the consumer products domain.
Authored deep-research prompts and designed multifaceted scoring rubrics for evaluating AI responses to consumer product queries. Developed factual-error detection, anti-hack criteria, and ensured tight rubric alignment with grounding URLs across multiple domains. Integrated multimodal tasks requiring image identification before text-based product analysis. • Created rubrics with 8–20 weighted criteria including grounding URLs and criterion tags. • Designed and validated fact-aligned, anti-hallucination, and negative scoring rubric components. • Developed self-contained, timeless evaluation criteria and prompt–rubric alignment checks. • Crafted prompts and tasks for both text- and image-based AI analysis in the consumer products domain.
2025 - Present