Limiting AI to Agents: Building Self-Learning Autonomous Systems with RAG, Agents, DSPy and MCP

[developerdhruv]

Title

Limiting AI to Agents: Building Self-Learning Autonomous Systems with RAG, Agents, DSPy and MCP

Describe your Talk

This session will begin with an exploration of Retrieval-Augmented Generation (RAG) and Agentic Systems, where we’ll delve into how knowledge retrieval enhances contextual intelligence, allowing AI systems to generate more informed and relevant responses. From there, we'll move into understanding how AI agents perform tasks, make decisions, and how to select the right frameworks—such as LangChain, CrewAI, AutoGen, and Pydantic AI—based on specific agentic goals.

Next, we’ll discuss building agents that can learn and adapt over time using language models. This segment will cover techniques for grounding knowledge in agents and enabling them to adjust their behavior as they interact with new data, ultimately fostering continuous learning.

The session will then shift to a hands-on introduction to DSPy, a framework that simplifies the process of composing language model programs. You'll learn how DSPy is making waves in the field and why it's gaining popularity in AI development. Alongside this, we'll also touch on the concept of Model Composition Programming (MCP) and how it enhances the versatility of language models in practical applications.

Finally, we’ll wrap up with a research-oriented discussion, diving into some of the latest papers and novel ideas in the realm of agent learning and self-optimization. This will give you a glimpse of cutting-edge developments in the field and how they are shaping the future of intelligent agents.

Pre-requisites & reading material

No response

Resources

No response

Time required for the talk

20-30 mins

Link to slides/demos

Slides
Explainer docs

About you

I'm Dhruv Kumar, a 16-year-old Software Engineer and AI Researcher from India, currently working as a Software Engineer at Samwell.ai (NY, USA). With 4 years of experience building AI-driven applications, I'm also a published researcher recognized by DST, India, and passionate about solving cross-domain problems through AI.

Availability

26 April

Any comments

No response

[Animesh-Ghosh]

Hey @developerdhruv , possible to share the public link to your slides?

[developerdhruv]

https://docs.google.com/document/d/1vxzgfAKIGaFK0J3dAWiXgbq9e5Fo-T1uFdInD5MGym8/edit?usp=sharing
Hi @Animesh-Ghosh sharing the core doc of session slides are in progress

[satyaakam]

@jaidevd do comment

[developerdhruv]

Hi @Animesh-Ghosh any update on this as event is too close.

[Animesh-Ghosh]

Hey @developerdhruv, sharing a few comments from @gora:

• Talk looks fine from a technical perspective
• Needs more of an introduction. Not clear what the focus of the talk is. What will the audience learn from the talk?
• Also suspect that it will go over the head of the average attendee, so you should include layman explanations as they go
• Might also be useful to mention the level of the talk (beginner, intermediate, advanced), and what prior knowledge is expected of the audience. Maybe share links for reading material

[agriyakhetarpal]

Hi @developerdhruv, would you like to deliver your talk in the May edition? While the venue and the logistics are undecided at the moment, it's slated to be on the 17th of May, and it could be in collaboration with PyData Delhi.

Please feel free to address the review points noted above and share a link to your slides whenever it's convenient for you to do so. Thank you!

[developerdhruv]

Yes sure, Sounds good!. Let me know the details in upcoming days.
Thanks.

[Animesh-Ghosh]

Yes sure, Sounds good!. Let me know the details in upcoming days. Thanks.

We'd suggest you to join our Telegram and/or WhatsApp groups. I have DMed both links over LinkedIn DM.

Please also update the proposal with link to your slides as soon as you can so that we can review the content.

Thanks!