Jessica Carr

Key Skills

Software

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Audio Recording

Freelancer Overview

Experienced researcher and AI training specialist with a PhD in Inclusive Citizen Science and a strong track record in LLM prompt engineering, usability research, and inclusive design. I have worked on projects involving the creation of PhD-level prompts to challenge LLMs across social science domains, developed evaluation rubrics for both text and image-based outputs, and authored audio-based prompts to test model interpretability. My work bridges rigorous academic research with practical, high-impact AI systems. At the Open University, I led inclusive EdTech initiatives such as the Student Dashboard and Digital Access Advisor, using participatory design and qualitative methods to improve AI-driven educational tools for disabled learners. I’ve collaborated with Microsoft, JISC, and Horizon Europe-funded teams, bringing together stakeholders to deliver ethical, user-centred solutions. With a background in accessibility, data governance, and UX research, I offer a rare mix of domain expertise and technical understanding ideal for AI training data work in education, accessibility, and human-centred design contexts.

Labeling Experience

Xylophone Spectrum

Scale AIAudioAudio Recording

Contributed to the development of an AI speech model by creating authentic, conversational audio prompts designed to simulate real human interactions. The goal was to help the model better understand and respond in natural, context-aware ways. This involved crafting prompts that reflected everyday speech patterns—as if speaking directly to another person—ensuring realism, clarity, and variation to support more effective and human-like AI responses.

Contributed to the development of an AI speech model by creating authentic, conversational audio prompts designed to simulate real human interactions. The goal was to help the model better understand and respond in natural, context-aware ways. This involved crafting prompts that reflected everyday speech patterns—as if speaking directly to another person—ensuring realism, clarity, and variation to support more effective and human-like AI responses.

2025

Pegasus

Scale AITextRLHF

Developed and solved advanced domain-specific problems designed to challenge and evaluate the limits of large language models (LLMs). The goal was to craft high-quality, complex prompts—particularly in the social sciences—that could "stump" the AI, revealing gaps in reasoning, nuance, or subject mastery. This work contributed to improving model robustness, fairness, and performance in real-world applications.

Developed and solved advanced domain-specific problems designed to challenge and evaluate the limits of large language models (LLMs). The goal was to craft high-quality, complex prompts—particularly in the social sciences—that could "stump" the AI, revealing gaps in reasoning, nuance, or subject mastery. This work contributed to improving model robustness, fairness, and performance in real-world applications.

2025 - 2025

Education

Work History