Building Robust, Secure and Scalable Agentic AI Applications

Topic: Building AI Agents

Agenda:

4:30 - 4:40 PM : Welcome Note

4:40 - 5:10 PM: Accelerating Agentic AI development with Model Catalog and Azure Open AI by - Avinash Bharadwaj

If 2024 was the year of AI Chatbots & RAG, then 2025 is the year of AI agents. While you may interact with an AI agent in the same manner you interact with an AI chatbot, the differences between them are stark. AI agents can autonomously act to fulfill your requests.

There are multiple approaches and frameworks out there to help one get started with Agentic AI applications, but more often than not, this can get complex and expensive. Thus, in this demo lead session we will explore open-source agentic AI tooling like Controflow to build Agentic AI applications using Agent Catalog and Azure OpenAI.

5:20 - 5:50 PM - Building Agentic Flows with Autogen Design Patterns by Aruna Chakkirala

In future humans will be working alongside Agents" that is the prevalent sentiment in the industry. Agentic Solutions are geared to be going mainstream soon. There are multiple means to building the Agents using no code or highly customizable pro code options which include Autogen and Semantic Kernel. This session will focus on Autogen for building Agentic Solutions. AutoGen is an open-source framework for building multi-agent systems with diverse roles and capabilities. Multi Agent Systems involve a combination of Intelligent Agents working together to achieve a goal. The collaboration patterns between the Agents can also be tuned to suit the objective and requirement.

This can be achieved by leveraging the multiple design patterns in Autogen which is well suited for implementing complex AI applications. In this session we will provide a background of Autogen through a demo and also delve into the multiple design patterns available to address scenarios

6 - 6:30 PM - Zero Trust Security for LLM/Agentic AI applications by Shivay Lamba

As businesses increasingly rely on LLM applications for their important functions, it becomes important to implement strong security measures to protect sensitive information and guarantee smooth operations. This session shares how to build a zero-trust security architecture for Agentic AI workloads using cloud native patterns. The session will explore how to implement AI Gateways that have strong authentication and authorization and include audit logging. Keep compliance and governance requirements while you secure model artifacts and implement runtime security and protect against prompt injection attacks.

6:30 - 6:40: Introduction to Semantic Kernel with Couchbase - Shivay Lamba

Quick introduction to Semantic Kernel for building agentic applications with the Couchbase integration

6:40 - 6:50: Implement AI Agents in your Developer Workflow using Github Copilot - Aditya Oberai

6:50 - 7:00 PM : Quiz, and closing note

7:00 - 8:00 PM - Refreshments and Networking

Location:
Microsoft Reactor Bengaluru
Vigyan”, No. 9, Ground Floor, Lavelle Road, Bengaluru, Karnataka 560001

Kindly take a note of the following Visitor Procedures:

• Visitors are requested to park their vehicles only at designated parking area as directed by the team and as per availability only.
• Yellow board vehicles are not permitted to park inside the premises.
• All visitors need to obtain “Visitors Badge” by verifying their government id like Aadhar card, Driving license or any other suitable Govt id at respective building reception and need to be escorted throughout their stay in Microsoft premises.
• Visitors should return the temporary visitor pass to the receptionist at the end of their visit. Visitors must abide by the "No Photography" & "No Smoking" policies on Microsoft premises