Freelance AI Model Trainer/Annotator
In this role, I trained and evaluated large language models (LLMs) in Hindi and English to enhance language understanding and generation. Tasks included fine-tuning through Reinforcement Learning from Human Feedback and providing granular ratings and rewrites to model outputs. I utilized diverse prompt types such as Rewrite, Summarization, Classification, Extraction, Closed QA, and more to test and strengthen AI model performance. • Contributed to projects like Cypher RLHF, Cypher i18 Evals, Bee SFT Multilingual, and Multilingual Annotations. • Rated and ranked AI model responses by evaluating instruction following, truthfulness, writing quality, harmlessness, and language fluency. • Made corrections and rewrites to outputs for better coherence, cultural, and contextual relevance in both Hindi and English. • Collaborated with remote teams to ensure high standards in multilingual AI training.