Kameshwari Kesavarapu

Google, Audi, Michelin

• Reviewed and annotated audio, video, images, and text to support machine learning applications, ensuring high-quality and accurate data labeling. Managed the entire annotation life cycle, from processing raw data to final dataset delivery, while adhering to project guidelines and quality standards.
• Developed and maintained data labeling guidelines to ensure consistency across multiple projects. Worked extensively on NLP tasks, including sentiment analysis, text classification, and intent detection, by labeling user reviews and accurately determining sentiment (positive, neutral, or negative). Ensured that annotations captured the intent and emotional tone of the text, including nuances like sarcasm, humor, and frustration.
• Conducted virtual agent testing, focusing on entity recognition, intent classification, and sentiment categorization. Created custom highlights, extracted key phrases, and integrated them into Google CCAI Insights. Designed and tested sample conversations to optimize Agent Assist performance.
• Collaborated with QA teams and conversational architects, supporting data-driven improvements and refining conversation flows in Dialogflow ES and CX to enhance chatbot and AI-assisted interactions.

Apple

• Trained junior associates in data annotation and NLP tasks, ensuring consistency and high-quality outputs across the team. Assisted in data preprocessing, including cleaning datasets, correcting errors, and structuring data for NLP applications.
• Annotated large-scale text datasets for sentiment analysis, labeling data as positive, neutral, or negative to enhance virtual assistant, SIRI. Worked with domain-specific datasets (e-commerce, media, entertainment, location, local search etc) to accurately tag sentiment and emotional intent, ensuring contextual relevance for the target audience.
• Participated in named entity recognition annotation, following NLP frameworks to improve AI-driven entity detection models. Designed custom ontologies, integrating relevant keywords to enhance AI model comprehension.
• Performed data validation and error checking, ensuring consistency and accuracy, particularly for complex sentiment expressions. Reviewed and refined translated content, applying cultural nuances for better clarity and contextual accuracy.
• Handled intent detection, entity tagging, sentiment annotation, semantic annotation, and speech-to-text processing, contributing to AI model training for improved virtual agent interactions.