utkarsh vaibhav

At Optum, I lead a 20-person Agentic Analytics organization building production systems for healthcare claims. I grew the team from 7 people, set the AI roadmap, and hired across data science, data engineering, and RPA. I shipped a LangGraph and Claude claims-rework pipeline that automates triage and recommendations across behavioral-health claims. It auto-resolves 25% of rework, reduced volume by 30%, and delivers more than $0.5M in annualized rework-cost savings. I also originated an enterprise Neo4j agentic knowledge graph across eight source systems, enabling same-day provider-risk detection and proactive education actions. Previously, I led ML systems for 4M+ daily UPI transactions at Accenture and built credit, retention, personalization, and forecasting models at Citigroup and Equifax. Across healthcare, fintech, and credit risk, I take ML from data engineering through production monitoring while building teams that turn analytics into measurable business outcomes.