AI Implementation & Digital Transformation Consultant – Data Labeling & Agent Training
Developed multi-agent AI simulation systems where agents perform actions based on processed financial documents and negotiation dialogues. Labeled and generated training data for these agents by annotating transaction text and user interaction logs, focusing on negotiation, payments, and loan approvals. Designed prompts and responses to improve AI agent understanding and loan decision transparency. • Responsible for annotating negotiation and financial decision dialogues in transaction logs. • Curated training/test datasets for LLM invoice-reading agents, optimizing prompt-response structures for realistic financial scenarios. • Led evaluation and feedback on AI output quality, iterating on ground truth labeling for fine-tuning models. • Applied fairness audits to labeled datasets, specifically targeting bias reduction for underserved groups.