Darsana Usha Devi

Title of the Talk:
From Data Platforms to Agentic AI: Operationalizing Enterprise GenAI at Retail Scale

Abstract :
Enterprise adoption of generative AI has moved beyond experimentation into the harder problem of operating it reliably at scale. Drawing on the journey of building a large retailer’s GenAI platform, this keynote examines what it takes to turn foundation models and agentic techniques into dependable, governed enterprise capabilities. It traces the path from traditional data and analytics platforms to a consolidated, MCP-based framework that unifies multiple AI initiatives — spanning retrieval-augmented generation, LLM orchestration, agent interoperability, and the evaluation and governance practices that make these systems trustworthy in production. The talk bridges industry practice with open research questions in computational intelligence and soft computing, highlighting where orchestration, human-in-the-loop design, and reliable evaluation remain unsolved. Attendees will leave with a practical view of enterprise GenAI architecture and a set of research directions grounded in real deployment challenges.

Bio:
Darsana Usha Devi is a Senior Manager of Product at Walmart, where she leads the development of enterprise generative AI platforms, including an MCP-based framework that consolidates multiple AI initiatives across the organization. With over 20 years of experience spanning AI/ML, data platforms, analytics, and retail, she has led enterprise-scale products in data monetization and first-party retail insights, and now focuses on operationalizing GenAI, RAG architectures, and LLM orchestration for real-world business impact. She is a Certified Scrum Product Owner (CSPO) and holds SAFe and Six Sigma Black Belt certifications. Beyond her product work, she is an active peer reviewer for international conferences and a hackathon judge, supporting emerging researchers and the broader AI community. She is based in Northwest Arkansas, USA.