NLP Sentiment Annotation Specialist (as part of AI/ML Engineer role)
Developed an NLP pipeline using BERT transformers to classify sentiment and intent from customer service transcripts. This involved annotating large datasets of customer queries for model training and evaluation, ensuring data quality and balanced class distribution. Regular audits and validation tasks were performed to improve model accuracy and reduce labeling errors. • Managed and oversaw classification tasks for conversation transcripts. • Conducted periodic quality checks and inter-annotator agreement assessments. • Collaborated with domain experts to refine label definitions and criteria. • Used internal/proprietary tools on Azure ML infrastructure for labeling workflow.