Agentic Systems with MCP + Visualizing Open Source Networks

Elastic and Confluent are happy to invite you to this exciting meetup in Vancouver!

Agenda:

• 6:30 pm: Doors open; say hi and eat some food
• 6:45 pm: Building Search-Aligned Autonomous Systems with MCP, Elastic, and Tron, by Justin Castilla, Sr. Developer Advocate at Elastic
• 7:15 Q&A with Justin
• 7:20 pm: Six Degrees of Open Source: How I’m mapping the OSS Collaboration Network, by Ryan Belgrave (Software Engineer at Confluent)
• 7:50 pm: Q&A with Ryan
• 8:00 pm: Networking and event wrap up

Abstracts:

Building Search-Aligned Autonomous Systems with MCP, Elastic, and Tron

Autonomous agents become more capable, yet they also become harder to trust - similar to the MCP of Tron long ago! The modern Model Context Protocol (MCP) introduces structure to an agent's reasoning by clearly defining context layers and transitions. In this talk, you'll learn how Elastic can serve as the memory, grounding, and trace layer for MCP-based systems—bringing observability and accuracy to otherwise “black-boxy” LLM behaviors.
We'll walk through a working example that shows how Elastic and MCP combine to build search-aligned, interpretable, and reliable agentic AI.Key Takeaways:

• Understand what MCP is and why it's important for agentic systems
• See how Elastic acts as both the memory and the observability layer
• Learn how to combine these for more robust and interpretable AI agents

Six Degrees of Open Source: How I’m mapping the OSS Collaboration Network

Inspired by the “Six Degrees of Kevin Bacon” concept, I’m exploring how to map connections within the Open Source community based on activity on Github. This session will demonstrate a method for visualizing the collaborative relationships between developers.
Here’s what I’ll be covering:

• Data Acquisition from Github: I’ll explain how to use the Github Events API to get information about various types of events happening on Github. I will go into detail about what events I am specifically interested in and what filtering needs to be done to get a good dataset.
• Data Processing with Bento and WarpStream: I’ll discuss writing a custom Bento plugin to read from the Github Events API, processing the data to filter out unwanted events, publishing the data into WarpStream, and writing Bento tests.
• Building a Collaboration Index with Elasticsearch: The processed data is then used to construct an index in Elasticsearch, which allows for visualizing and analyzing the connections between developers through search and aggregations.
• The benefits of WarpStream: I’ll explain the technical reasons for using WarpStream for a project like this and how its architecture helped minimize costs.

If you are interested in stream processing, distributed systems, or interested in how many degrees of separation you are between various open source contributors this session is not one to miss!

📅 Date: September 9th
⏰ Time: 6:30 PM
📍 Location: UBC Robson Square, Room C.400
The University of British Columbia
800 Robson St, Vancouver, BC V6Z 3B7

See you there!

Cheers!