​AI demos are everywhere.

​But how do companies actually build AI products that work reliably with live data, low latency, and thousands of users?

​Join engineers as they share the architectures behind production AI systems - from streaming data and event-driven architectures to real-time databases, RAG pipelines, observability, and open-source infrastructure.

​No hype. Just practical engineering stories, lessons learned, and opportunities to meet others building modern data systems.

​Food and drinks included 🍻

​

​PS. Want to hear more great speakers? Sign up for the Open Source Analytics and AI Conference on Nov 2 (online and in-person). Register here.

​

​Agenda

• ​6 pm - Networking

• ​6:15 - 8:00 pm - Talks

• ​8:00 - 9:00 pm - Networking

​Speakers

• ​Robert Hodges, CEO @ Altinity

• ​Goutham Veeramachaneni, Staff Software Engineer @ Grafana

• ​Aashish Nair @ dltHub

​Description of talks

​Tools and Tricks to Prepare Your ClickHouse® for AI

​Speaker: Robert Hodges, CEO @ Altinity

​Abstract: AI is coming! It's going to crush your ClickHouse database!!  Well maybe. This talk will cut through the FUD, focusing on three ground-level issues that are emerging across a wide range of installations. First, build sandboxes to protect data if (when) your agents go rogue. Second, lower costs of operating ClickHouse to free up resources for AI. Third, use AI models to diagnose problems and guide application improvements. We're working on all three at Altinity and will share lessons on solving them.

​

​Vibe Building AI Applications with Confidence

​Speaker: Goutham Veeramachaneni, Staff Software Engineer @ Grafana

​Abstract: I am no AI engineer, however, I am building a lot of small home-cooked apps for my specific use-cases. I also am at a point that I don't read the code anymore for these (hides shame). The only saving grace here is that I ensure proper observability is baked in from the get-go and continuously use the observability data and then use the observability data to continuously improve the applications.

This practice is borrowed from how we build our Grafana Assistant and we'll dive into a practical demo of how I improved one of my LLM powered apps, and then share how you can do the same for yours.

​

​Ingesting Agent Traces with dltHub

​Speaker: Aashish Nair, GTM Engineer @ dltHub

​Abstract: Every production agent leaves a trail: traces, tool calls, errors. Most of it sits wherever your observability tool caught it first, so nobody can tell you what your agents cost at scale or which ones are quietly failing. To answer that, traces need to land in your warehouse next to product observability, Anthropic API costs, and the rest, which means someone has to build and run a pipeline. Writing that code was never the hard part. The hard part is schema drift, silent failures, and the API that breaks at 3am on a Saturday.

​dltHub closes that gap as the agentic data and managed-infra layer on top of your warehouse: a coding agent builds, runs, and fixes pipelines with good data engineering practice enforced, and managed infrastructure keeps them alive in production.

​Aashish shows it live: prompting an AI agent to build a pipeline that pulls traces from dltHub's own in-production chatbot into ClickHouse, then deploying it on dltHub to run and maintain itself.

​

​​Follow the OSA Community

​Make OSA COM your community.

​Website: www.osacom.io

​LinkedIn: /osa-con

​Slack: https://go.osacom.io/slack