AI training / red-teaming and evaluation contributor for multiple platforms (Outlier, Alignerr, Lionbridge, Babel Audio)
Provided AI training and evaluation support across multiple external AI platforms, including Outlier, Alignerr, Lionbridge, and Babel Audio. Performed red-teaming, RLHF-related tasks, and rubric testing to help improve model behavior and response quality. Conducted side-by-side reviews to compare outputs and support iterative dataset/model improvement. • Red-teaming tasks for safety and policy compliance • RLHF data work and rubric-driven evaluation • Side-by-side reviews of model responses • Rubrics testing and generation