AI & LLM Data Analyst | Innodata India
I performed daily analysis and labeling of hundreds of multimodal data samples to directly support large language model (LLM) training and evaluation. My primary responsibility involved ensuring high data quality through structured reviews and annotating data with over 97% accuracy. I also delivered reinforcement learning from human feedback (RLHF) evaluations and generated prompt analysis reports to flag model issues. • Labeled and reviewed at least 500 multimodal data samples every day. • Conducted consistency checks and helped reduce annotation rework by approximately 20%. • Composed reports highlighting edge cases and ensuring improved alignment of AI outputs. • Collaborated on the development and documentation of QA standards for a multidisciplinary analytics team.