Vinoth Premkumar

Sephora | Trend Intelligence Application

• Architected autonomous multi-agent workflow using Google ADK that automates end-to-end trend discovery and competitive benchmarking, achieving 85% faster turnaround than manual research processes
• Built scalable "Trend Lake" on Databricks Delta Lake persisting multi-modal data (social metrics, expert commentary, images) with competitor analysis engine that identifies market gaps and trending ingredients in real-time
• Delivered executive dashboard with real-time trend visuals and sentiment analytics, directly accelerating product launch decision-making for business stakeholders

PepsiCo | Dialogue with Data

• Developed graph-based revenue analytics platform using LangGraph and Neo4j that enables natural language queries on complex financial data, eliminating technical barriers for non-technical category managers to analyze price elasticity, revenue optimization, and assortment strategies
• Automated metadata orchestration with specialized agents that autonomously identify and map relevant data assets across disparate SQL tables, transforming sophisticated analytical questions into actionable insights with 60% query accuracy

PepsiCo | AI Migration Assistant for Data Modernization

• Led AI-driven migration from legacy Oracle/Informatica to Databricks supply chain environment using intelligent multi-agent SQL parsing, reducing manual source-to-target mapping effort by 40% and accelerating Medallion architecture implementation
• Engineered autonomous parser and Neo4j lineage visualization that analyzed legacy PRO files and Informatica XMLs, providing clear migration roadmap and ensuring high-fidelity pipeline rewrites for critical supply chain planning applications