Multilingual AI Training & Annotation for Generative Model Development
I contributed to a large-scale generative AI training initiative, progressing from data annotator to Verified Advanced AI Trainer and Fix-It Specialist. My work involved labeling and categorizing multilingual text data (English, Arabic, Urdu, Turkish), crafting prompts and ideal responses for supervised fine-tuning (SFT), and evaluating model outputs using RLHF techniques. I performed entity recognition, classification, emotion tagging, and scenario-based dialogue evaluation to improve model alignment and reasoning. As a Fix-It Trainer, I audited and corrected other trainers’ work to ensure flawless, client-ready outputs. The project spanned thousands of data points and required precision, consistency, and cross-functional collaboration with technical and editorial teams.