Michael Wells

I am currently an AI data annotator working as a Generalist, and I’m looking to expand my opportunities and options in this field. I focus on evaluation work that improves how AI model outputs are judged for quality, accuracy, and usefulness.

My day-to-day responsibilities include comparison-based AI model response evaluation using rating axes and written explanations for SFT, RLHF, and red-teaming purposes. I also research factual claims made by AI models to assess their currency and accuracy.

I perform heuristic checks for prompt fulfillment, relevance, verbosity, scannability, style, and tone. I also research the accuracy of images in educational widgets, looking for anomalies and artifacts, and I handle fact-checking, proofreading, and editing text.

Beyond annotation, I create prompts and rubrics to test models’ capabilities and limitations, and I evaluate co-workers’ task submissions to ensure alignment between ratings and rationale. I bring additional domain expertise from 42 years of experience as a musician specializing in guitar, plus experience writing college papers, poetry, short stories, and technical writing centered on music and guitar instruction.