At Amazon, I oversee data quality operations for Generative AI annotation workflows, identifying edge-case inaccuracies and factual hallucinations in training data. I also analyze data trends and establish quality metric dashboards to support reliable model evaluation.
As an ML Data Associate - II at Amazon, I supervised annotation and RLHF pipelines across multi-modal datasets and evaluated prompt-response pairs for quality, safety, and accuracy. I helped launch 10+ operational solutions across sites, achieving 90% of targeted client and business performance goals.
Earlier, I delivered labeled text and speech data for ML training at Amazon and managed catalogue data at Wissend Technologies. I also conducted root cause analyses for an AHT optimization program that reduced Average Handling Time by 50% without compromising data quality.
