Rakesh Aggarwal

Inside the Architecture of Enterprise GenAI Applications

Submitted Dec 16, 2025

Session Overview:
Modern Enterprise GenAI applications require far more than simply integrating an LLM or deploying agents. They demand a thoughtfully engineered, layered architecture that enables observability, performance optimization, security, and controlled automation at every stage of execution.

In this session, we will explore how a production-grade Enterprise GenAI system can be designed using a layered architectural model, where each layer plays a distinct role and is continuously analyzed and observed to ensure reliability and scalability.

The session will break down the following architectural layers:

Interface Layer — APIs, chat interfaces, and UI components that translate human intent into structured machine-executable inputs

Reasoning / Orchestration Layer — the “brain” of the system, responsible for agent planning, workflow decomposition, tool selection, and decision-making

Tools & Integrations Layer — real-world interactions with APIs, databases, enterprise platforms, and third-party systems

Memory Layer — short-term and long-term memory enabling contextual recall, personalization, learning, and state management

Action Layer — closed feedback loops where AI-driven decisions update backend systems and trigger automated workflows

The session emphasizes how observability, memory design, and orchestration strategy dramatically influence the effectiveness of GenAI solutions in real enterprise environments.

Key Takeaways:
Understand why LLMs and agents alone are not sufficient for enterprise-grade GenAI systems

Learn how architectural decisions impact performance, cost, reliability, and governance

See how memory, orchestration, and integrations work together to enable real-world automation

Gain clarity on how organizations can move beyond proofs-of-concept into stable, secure, and scalable GenAI deployments

This approach combines practical engineering discipline with hands-on AI implementation, helping businesses bridge the gap between experimentation and production readiness.

Target Audience:
This session is designed for:
CIOs, CTOs, and Enterprise Architects
Leaders shaping AI strategy who want to understand how GenAI architectures differ from traditional application designs and what changes are required at scale.

ML Engineers, Backend Developers, and DevOps Engineers
Practitioners navigating rapidly evolving AI technologies and looking to understand how GenAI fits into modern engineering workflows—without the hype or FOMO.

Learners and Software Engineers Exploring GenAI
Professionals who perceive GenAI as overly complex and want clarity on how existing software engineering skills translate into AI-driven system design.

About the Speaker:
Rakesh Aggarwal brings over 13 years of experience across engineering, quantitative analysis, e-commerce platforms, ITSM systems, Salesforce ecosystems, DevOps, security, and GenAI implementations.

With deep, hands-on exposure to designing and delivering enterprise systems, Rakesh offers a grounded, real-world perspective on Agentic frameworks and enterprise GenAI architecture. His sessions focus on clarity, practical design patterns, and lessons learned from production environments—helping attendees understand not just what to build, but how to build it right.

Comments

{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Hosted by

Jumpstart better data engineering and AI futures