IN PERSON Apache Kafka® Meetup Chennai - May 2025

Hello everyone! Join us for an IN PERSON Apache Kafka® meetup on May 31st from 9:30 am.

📍 Venue:
Saama Technologies, OLYMPIA TECH PARK, Chennai

Agenda:

• 09:30 AM - 10:00 AM | Registration and Welcome
• 10:00 AM - 10:45 AM | KAN: Making Neural Networks Interpretable and Adaptable
• 10:45 AM - 11:00 AM | Break & Networking
• 11:00 AM - 11:45 AM | Vector embeddings and similarity search with Flink SQL
• 11:45 AM - 12:30 PM | Kafka Meets the Model Context Protocol
• 12:30 PM - 01:30 PM | Lunch & Networking

Session 01:
💡 Speaker :
Selvakumar Murugan, Senior Principal Research Engineer, Saama

Talk:
KAN: Making Neural Networks Interpretable and Adaptable

Abstract:
- What if we ditched activation functions in neural nets altogether?
- That is exactly what, Kolmogorov–Arnold Networks (KANs) do -- replacing them with learnable splines on the edges.
- This small change results in better interpretability, smoother function fitting, and often fewer parameters.
- KANs show strong results in scientific computing, symbolic regression, and even PDE solving.
- We’ll break down how they work, where they shine, and why they matter.
- If you're curious about the next evolution in AI models, this session is for you.

Session 02:

💡 Speaker :
Diptiman Raichaudhuri, Staff Developer Advocate, Confluent

Talk:
Vector embeddings and similarity search with Flink SQL.

Abstract:
- Vector embeddings and similarity search on embeddings have unlocked AI use cases which were previously not used to derive insights from unstructured data.
- This has opened up new frontiers around RAG and with modern techniques, such similarity searches could also be performed on real-time streaming data.In this session, an introduction to vector embeddings would be presented.
- Taking the audience through different intuitions based on which modern embedding models are created. The evolution of embedding techniques from TF-IDF to word2vec to Transformers.
- The last section of this session would introduce how embeddings are changing the real-time streaming landscape.
- This would be demonstrated with Flink invoking a remote LLM and introducing similarity search with a Kafka-Flink data streaming pipeline.

Session 03:
💡 Speaker :
Lakshmi Narasimhan Parthasarathy, CTO, Kubenest

Talk:
Kafka Meets the Model Context Protocol

Abstract:
- Discover how the Model Context Protocol (MCP) is transforming AI integration by providing a universal, open standard for connecting large language models with external tools, APIs, and data sources-eliminating the need for brittle, custom integrations.
- Learn how combining MCP with Apache Kafka’s real-time event streaming capabilities enables scalable, modular, and robust AI-powered applications that can interact seamlessly with complex enterprise systems.
- This talk will explore use cases for leveraging both Kafka and MCP to build the next generation of intelligent, context-aware solutions.

***
If you would like to speak or host our next event please let us know! community@confluent.io