WhatsApp Chatbot AI Data Labeling and Entity Extraction (GTT Data Solutions)
Worked on automating donor workflows with intelligent intent classification on WhatsApp chat data for an NGO using AI. Developed and deployed entity extraction and multi-turn conversation AI to classify intent, extract entities like payment details, and drive downstream workflow automation. Leveraged OCR and natural language understanding to structure large volumes of unstructured chat inputs for use in ERP, ticket routing, and KYC operations. • Labeled and classified 50+ intent categories in donor conversations, developing robust entity extraction techniques for PAN, Aadhar, addresses, and more. • Annotated data for training AI models and conversational engines, including conversations in multiple languages (Hindi Devanagari/Latin, English, Hinglish). • Used Google Cloud Vertex AI and Gemini AI to build, label, and validate entity recognition models. • Improved data completeness and automation through structured annotation and model feedback cycles.