CUA Tutor
I managed the post-training and alignment of coding and reasoning models by curating high-quality human preference data. I designed and implemented human data generation pipelines focused on computer user agents, optimizing for autonomous interaction. My responsibilities included leading annotation, validation, and QA processes, enhancing both model performance and real-world impact. • Curated and generated labeled data for coding models • Oversaw large-scale annotation and validation workflows • Applied rigorous quality assurance for human-labeled datasets • Directed pipeline design for improved tool-use modeling