AI Infrastructure Night
About the event
AI Infrastructure Night brings together founders and engineers building the systems behind production AI.
Five short talks will cover software deployed into customer clouds, event-driven infrastructure for agent traffic, GPU-fleet reliability, and the practical work of technical founders-followed by networking with the people building these systems.
🎟️ Space is limited. Request an invite to save your spot.
✨ Speaker lineup
5:30–6:00 PM: Arrival, food & networking
6:00pm - Tom Hacohen, Founder & CEO, Svix
Agents are going to kill your service
Agent load is coming for everyone's servers, and without preparation they will lead to significant cost increases, complexity, and scaling challenges. In this talk Tom will talk about the alternative: event-driven architecture, the main challenges with adopting it, how to solve these challenges, and how make sure your infrastructure is agent-ready before it falls flat on its face.
6:15pm - Alon Gubkin, Founder & CEO, Alien
How Do You Operate Software Inside Infrastructure You Don’t Control?
Deploying into a customer’s cloud once is straightforward; upgrading and debugging it across hundreds of opinionated environments is not. This talk covers drift, permissions, partial failures, upgrades, rollbacks, observability, and supporting Terraform, Helm, and internal platform tooling without direct access.
6:30 pm - Roi Talpaz, Co-Founder & CEO, Backdrop
Stop Pitching: How Technical Founders Should Run a Discovery Call
Technical founders default to explaining the product. Buyers default to tuning out. This talk covers the question-led structure that wins first calls, the three questions that separate a real deal from a polite conversation, and why the best discovery calls have the founder talking less than a third of the time.
6:45pm- Aditya Nath, Member of Technical Staff, Fireworks AI
Why Kubernetes Node Health Fails in Heterogeneous GPU Fleets
In AI inference, a cluster can report 100% Ready while silently hemorrhaging throughput because AWS, GCP, and alternative GPU clouds ship mismatched kernels, drivers, and fabric topologies. We’ll explore why standard kubelet checks and dashboards miss cross-cloud drift, and how to use immutable runtime recipes to catch performance degradation before inference SLOs take the hit.
7:00pm - Gregory Koberger, Founder, ReadMe
Your API is now your product APIs used to primarily be an integration point for code. Now, they're becoming the main access point for your product. Greg will talk about how to make sure your API is ready for a world where the main way people interact with you is via APIs and MCP servers.
7:15 PM onward — Networking