Sri Venkata Aravindbabu Malempati

Agentic AI: Autonomous Intelligence for Complex Goal Achievement

Abstract:

The rapid advancement of Large Language Models and their underlying transformer architectures has triggered one of the most consequential shifts in AI history, moving us from systems that answer questions to systems that autonomously achieve goals. Yet most organizations remain unprepared for the architectural and operational changes this demands.

This talk provides a rigorous, practitioner-focused framework for understanding agentic AI: autonomous computational systems that iteratively perceive their environment, reason across complex problem spaces, formulate multi-step plans, execute actions via tools, and self-correct based on feedback, all with minimal human intervention. Unlike traditional ML systems that learn patterns from data, or rule-based systems constrained by explicit programming, agentic AI combines both with proactive, goal-directed autonomy.

We examine the five core architectural components that separate true agentic systems from simple LLM wrappers: perception, reasoning, planning, tool use, and reflection. We explore how systems like Claude Code autonomously write, test, debug, and deploy production software, accessing development environments, executing code, running test suites, and iterating on implementations to resolve failures. In enterprise settings, agentic systems are automating complex workflows, including project management, customer support, scheduling, document processing, and data analysis that previously demanded significant human effort.

A central challenge explored is memory architecture: how agentic systems manage four distinct memory types, short-term working memory, episodic memory, semantic memory, and procedural memory, within the finite context windows of today's LLMs, a constraint that directly limits reliability at scale and remains an active area of research.

Attendees will leave with a clear definitional framework, a practical lens for evaluating agentic system readiness, and a grounded view of where the technology's frontiers and risks currently lie.

Profile:

Sri Venkata Aravind Babu Malempati is a Java Full Stack Principal Engineer with over 14 years of experience in analysis, design, development, architecture, and implementation of enterprise applications across banking, telecommunications, and insurance industries. He is based in St. Louis, MO.

Aravind has deep expertise in Java/J2EE, Spring Boot, and microservices architecture, with hands-on experience across AWS and Azure platforms. His technical stack includes Angular, ReactJS, Node.js, Python, and Golang, backed by strong skills in Hibernate, Kafka, Apache Flink, and Docker/Kubernetes for containerized deployments.

At Capital One Bank, he led the development of microservices supporting credit card autopay and auto loan recurring payment systems across the US and Canada. At Verizon, he designed and developed APIs using Java and Spring Boot for a Device Payment Agreement program enabling customer installment payments. At MUFG Bank, he built cloud microservices and REST web services supporting identity and access management functions. At Assurant Inc., he migrated legacy projects to Spring Boot, Spring Data, and Spring Cloud while developing a global self-service portal for claims and policy management.

Aravind works within Agile and Scrum environments and practices test-driven and behavior-driven development. He has experience with CI/CD pipelines using Jenkins and Git, and has maintained Splunk environments for monitoring and documentation.

He holds a Master's degree in Engineering and Management from California State University and a Bachelor's degree in Electronics and Communication Engineering from JNTU College of Engineering, India.