Bhargavi Kalicheti

Beyond Speech-to-Text: Voice-to-Voice Generative AI for Emotion-Aware Healthcare Contact Centers

Abstract:

Healthcare insurance organizations handle millions of member, provider, and pharmacy interactions through telephony channels each year, supporting tasks such as eligibility verification, benefit inquiries, claims clarification, prior authorization status checks, and document access. Despite the growth of digital self-service, voice remains the primary interface for complex and time-sensitive healthcare inquiries. However, most contact centers still rely on traditional interactive voice response (IVR) systems built on deterministic decision trees and rule-based call flows. These systems struggle to interpret natural language, retain conversational context, and adapt to evolving policies, resulting in frequent misrouted calls, repeated agent transfers, and increased operational burden for large healthcare contact centers.

This talk presents a reference architecture for an end-to-end conversational Voice AI ecosystem designed to enable real-time telephony intelligence in regulated healthcare environments. The architecture integrates large language models (LLMs), natural language understanding (NLU), conversational orchestration layers, and enterprise workflow integrations within a cloud-native microservices framework. The proposed ecosystem decomposes the conversational pipeline into independently scalable services, including speech-to-text ingestion, intent reasoning, knowledge retrieval, generative response planning, and text-to-speech synthesis, allowing each component to scale horizontally based on workload signals.

A key architectural contribution is the hybrid reasoning model, which combines lightweight intent detection models for rapid classification with LLM-based reasoning for complex or ambiguous requests. Generative outputs are grounded through policy validation layers, predefined response schemas, and structured execution plans, ensuring that conversational flexibility is balanced with deterministic behavior and regulatory compliance.

The architecture also incorporates stateless conversational processing with externalized state management, enabling elastic scaling, fault isolation, and multi-region deployment for high-concurrency environments. Additional layers for observability, auditability, and governance provide explainable routing decisions and full traceability of conversational outcomes critical requirements in regulated healthcare systems.

By combining cloud-native scalability, governed generative reasoning, and enterprise workflow grounding, this architecture provides a production-ready blueprint for deploying conversational Voice AI at a national scale, enabling healthcare organizations to modernize telephony systems, improve self-service effectiveness, and support increasingly complex member interactions.

Profile:

Bhargavi Kalicheti is a technology leader with over 15 years of experience designing and delivering enterprise-scale solutions across artificial intelligence, machine learning, healthcare technology, and large-scale IT systems. She specializes in developing AI-powered conversational platforms, Generative AI solutions, and agentic AI architectures that enhance healthcare service delivery and support large-scale digital transformation initiatives.

Currently serving as a Lead AI/ML Engineer at UnitedHealth Group (Optum), Bhargavi leads the modernization of legacy telephony and IVR ecosystems into AI-enabled, cloud-native conversational platforms that support interactions among members, providers, and pharmacies. Her work focuses on building scalable AI systems using technologies such as large language models (LLMs), prompt engineering, retrieval-augmented generation (RAG), and autonomous AI agents. These solutions enable natural, context-aware interactions and help organizations improve operational efficiency while maintaining regulatory compliance in healthcare environments.

Bhargavi has played a key role in designing multi-agent AI systems that plan, reason, and execute complex workflows across healthcare platforms. These systems integrate enterprise APIs, knowledge bases, and decision engines to support tasks such as prescription status checks, benefits validation, and secure data retrieval. She has also implemented guardrails, monitoring frameworks, and human-in-the-loop controls to ensure safe and responsible deployment of AI technologies in regulated environments.

Throughout her career, Bhargavi has delivered several high-impact initiatives that improved operational efficiency and reduced costs. Her AI-driven conversational platforms have increased call containment, reduced contact center volumes, and improved provider and member experiences across healthcare systems. These solutions support thousands of user interactions and generate measurable operational savings by improving automation and self-service capabilities.

Before joining UnitedHealth Group, Bhargavi worked as a Technology Analyst at Infosys Technologies, where she contributed to the development of enterprise applications, modernized legacy systems, and built scalable microservices architectures that later enabled AI and digital platform expansion.

Bhargavi holds a Postgraduate Program in Artificial Intelligence and Machine Learning from the University of Texas at Austin and a Bachelor of Technology in Information Technology from SSN College of Engineering, Chennai. She also maintains professional certifications, including AWS Certified Solutions Architect – Associate, Google Cloud Digital Leader, Sun Certified Java Developer, and Certified SAFe DevOps Practitioner.