Hands On | OpenAI Agents SDK

[Arko-Sengupta]

Talk title

Build Quick Agents with OpenAI Agent SDK

Short talk description

Discover How To Build Intelligent AI Agents In Minutes Using OpenAI's Agent SDK. This Talk Demystifies The Process Of Creating Autonomous Agents That Can Reason, Plan, And Execute Complex Tasks. Learn The Core Concepts Of Agent Architecture, Tool Integration, And Multi-Step Workflows. We'll Walk Through Practical Examples—From Simple Task Automation To Advanced Multi-Agent Systems. Whether You're Building Customer Support Bots, Research Assistants, Or Workflow Automation Tools, This Session Will Equip You With The Knowledge To Rapidly Prototype And Deploy AI Agents. Perfect For Developers Looking To Leverage The Latest In Agentic AI Without Getting Lost In Complexity.

Long talk description

AI Agents Are Revolutionizing How We Build Intelligent Applications, And OpenAI's Agent SDK Makes This Technology Accessible To Every Developer. This Talk Provides A Comprehensive Guide To Building Production-Ready AI Agents Quickly And Efficiently.

We'll Start By Exploring What Makes An AI Agent Different From Traditional Chatbots—Understanding Concepts Like Reasoning, Planning, Tool Use, And Memory. You'll Learn How Agents Can Break Down Complex Problems Into Manageable Steps, Make Decisions, And Interact With External Systems To Accomplish Goals Autonomously.

The Session Covers The Essential Building Blocks Of The OpenAI Agent SDK: Setting Up Your Development Environment, Configuring Agent Behavior, Implementing Custom Tools And Functions, And Managing Multi-Turn Conversations With Context Retention. We'll Dive Into Real-World Implementation Patterns Including Error Handling, Streaming Responses, And Optimizing Agent Performance For Production Use.

We'll Also Address Important Considerations Like Cost Management, Rate Limiting, Security Implications Of Autonomous Systems, And Testing Strategies For Non-Deterministic Agent Behavior. You'll Leave With A Clear Understanding Of When To Use Agents Versus Traditional Approaches, And How To Architect Scalable Agent-Based Solutions.

Whether You're A Backend Developer Exploring AI Integration, A Full-Stack Engineer Building Intelligent Features, Or An AI Enthusiast Ready To Move Beyond Simple LLM Calls, This Talk Will Accelerate Your Journey Into Agentic AI Development.

What format do you have in mind?

Talk (20-25 minutes + Q&A)

Talk outline / Agenda

1. Introduction To AI Agents

• What Are AI Agents Vs. Traditional Chatbots?
• Key Capabilities: Reasoning, Planning, Tool Use, Memory
• Real-World Use Cases And Applications

2. OpenAI Agent SDK Fundamentals

• Architecture Overview And Core Components
• Setting Up The Development Environment
• Basic Agent Configuration And Initialization
• Understanding The Agent Execution Loop
• Tool Integration And Function Calling

3. Building A Practical Agent - Live Demo

• Step-By-Step: Personal Assistant Agent
• Defining Tools And Custom Functions
• Managing Context And Multi-Turn Conversations

4. Advanced Patterns And Best Practices

• Multi-Agent Collaboration Concepts
• Prompt Engineering For Optimal Agent Behavior
• Memory Management And State Persistence
• Performance Optimization Techniques

5. Production Considerations

• Cost Management And Rate Limiting
• Security And Safety Guardrails

Key takeaways

1. Understand The Fundamental Difference Between AI Agents And Traditional Chatbots, Including Core Concepts Like Reasoning, Planning, Tool Use, And Memory Management That Enable Autonomous Task Execution.

2. Gain Hands-On Knowledge Of The OpenAI Agent SDK Architecture And Learn How To Set Up, Configure, And Initialize Agents For Production-Ready Applications.

3. Master The Process Of Building Custom Tools And Functions That Allow Agents To Interact With External APIs, Databases, And Services To Accomplish Complex Multi-Step Tasks.

4. Learn Best Practices For Agent Development Including Prompt Engineering Techniques, Error Handling Strategies, Context Management, And Debugging Non-Deterministic Agent Behavior.

What domain would you say your talk falls under?

Data Science and Machine Learning

Duration (including Q&A)

30 Minutes (20 Minutes Session + 10 Minutes Q&A)

Prerequisites and preparation

• Basic Python Programming: Familiarity with Core Python Fundamentals.
• API Understanding: Basic Knowledge of RESTful APIs and HTTP Requests.
• Experience With LLMs: Prior Exposure to Working with ChatGPT, GPT-4, or Similar Language Models is Helpful but not Mandatory.
• Async Programming: Understanding of Asynchronous Programming Concepts is beneficial but will be Explain during the Talk.

Resources and references

No response

Link to slides/demos (if available)

Please Note that the Provied Content is just an overview or introduction that we'll gonna. More enhancements and productive content is in progress.

Drive URL: https://docs.google.com/presentation/d/1AA13pPXVUtY1wdc3anda5o-nfCTLQXIU/edit?usp=sharing&ouid=100250443884081515040&rtpof=true&sd=true

Twitter/X handle (optional)

https://x.com/_Arko_Sengupta_

LinkedIn profile (optional)

https://www.linkedin.com/in/arko-sengupta-p/

Profile picture URL (optional)

https://drive.google.com/file/d/1WQ21KtDfVF6iOutI5ef3VnHO2avga-Am/view?usp=sharing

Speaker bio

Hey there,

I'm Arko Sengupta a Machine Learning Engineer at Appinventiv With A Passion For Building Intelligent Applications Using Cutting-Edge AI Technologies. With 3+ years Of Experience in Data Science & Machine Learning.

Driven By A Belief That AI Should Be Accessible To Every Developer, I mostly Focus On Demystifying Complex Concepts And Providing Practical, Hands-On Guidance. When Not Coding, I Enjoys Singing and Playing Guitar, And Actively Participates In Local Tech Community/Meetups/Hackathons.

Connect With me On LinkedIn and XCorp

Availability

I'll be Available for the 8th November Tech Session.

Accessibility & special requirements

• High Resolution Display Setup
• Reliable Internet Connection

Speaker checklist

• I have read and understood the PyDelhi guidelines for submitting proposals and giving talks
• I will make my talk accessible to all attendees and will proactively ask for any accommodations or special requirements I might need
• I agree to share slides, code snippets, and other materials used during the talk with the community
• I will follow PyDelhi's Code of Conduct and maintain a welcoming, inclusive environment throughout my participation
• I understand that PyDelhi meetups are community-centric events focused on learning, knowledge sharing, and networking, and I will respect this ethos by not using this platform for self-promotion or hiring pitches during my presentation, unless explicitly invited to do so by means of a sponsorship or similar arrangement
• If the talk is recorded by the PyDelhi team, I grant permission to release the video on PyDelhi's YouTube channel under the CC-BY-4.0 license, or a different license of my choosing if I am specifying it in my proposal or with the materials I share

Additional comments

No response

[Animesh-Ghosh]

Hi @Arko-Sengupta, thanks for your proposal!

The session outline seems fine, could you update the original comment with the slides for your session? They don't have to be the final slides fyi, just something we can review and give feedback on.

Thanks!

[Arko-Sengupta]

Hi @Arko-Sengupta, thanks for your proposal!

The session outline seems fine, could you update the original comment with the slides for your session? They don't have to be the final slides fyi, just something we can review and give feedback on.

Thanks!

Hey @Animesh-Ghosh I've Done the Same. Let me know if anything else is required from my end.

[Animesh-Ghosh]

The slides look good @Arko-Sengupta.

It wasn't clear from the slides, however, if there would be a single agentic app, sort of like a capstone project, that the audience would be building as part of the session. If there isn't one, we could try and push the session within a 30 minute slot.

[Arko-Sengupta]

@Animesh-Ghosh Yes, I think that's a Good Idea. As OpenAI is not a Open Source Tool therefore it would be difficult to get the key for everyone. Work for me! Let's keep it 30 Mins.