/ About this event
Infrastructure, ML, and application development are converging fast—and observability is where they meet. We're bringing together the people building across all three to talk about what changes when you can see everything.
Boba all night, snacks throughout.
Agenda
[6:00] Doors, networking, and boba [7:00] · The Deconstructed Database at Datadog — Julien Le Dem, Principal Engineer at Datadog
The Deconstructed Database at Datadog Speaker: Julien Le Dem, Principal Engineer at Datadog
As Datadog scaled to handle over a hundred trillion daily events, our engineering teams built their own custom query engines to keep things fast. While this helped teams move quickly, it also created silos that made it tough to analyze data across different products. To keep growing and support AI, we had to find a way to bring these engines together using open standards without slowing anyone down or risking reliability. In this session, Principal Engineer Julien Le Dem shares how Datadog turned its query engines into a more flexible, composable system. He'll discuss how we're defining clear contracts that improve performance and give teams autonomy, while breaking down those old silos. He'll also walk through how we're using and contributing to major open-source technologies to build it. You'll walk away with insights on how you can use modern open standards to make your own purpose-built engines more flexible and resilient, even at a massive scale.
Kubernetes Is Not Your Sandbox Speaker: Mislav Ivanda, DevEx Engineer at Daytona
Teams are reaching for Kubernetes to run agent sandboxes, and it's the wrong tool. Kubernetes is built to keep things alive and hold them in a steady state. A sandbox is born, forked, and killed before any of that machinery catches up.
The mismatch compounds because the sandbox keeps gaining requirements without shedding any. In eighteen months it went from a fast code-snippet runner, to a stateful box for long-running agents, to ten thousand ephemeral environments that fork for RL rollouts and die in under a second. It has to be all of those at once, a contradiction set no orchestrator was designed to hold. The cost shows up the moment you measure it. We ran the same 50-action bug-fix trajectory across five stacks and got a 12x spread: 12.9s on the fastest, 161.5s on the slowest. The gap isn't compute, it's lifecycle overhead per action. We name every stack and explain the mechanism behind each number.
[More speakers loading...]
Hosted by Datadog
This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com.
