Task attempter and contributor, Scale AI (AI training and evaluation)
Worked on AI training projects by reviewing responses from two AI model variants and selecting which output was more accurate, natural, and helpful. Helped implement training by performing onboarding and assessments for the project workflow. Produced human-judgment ratings used to improve model performance. • Reviewed and compared model A vs model B outputs • Rated outputs and generated preference judgments for training data • Corrected model mistakes based on evaluation results • Wrote and revised prompt-related sample answers/summaries to improve quality