Safety & Velocity: From AI-Assisted Dev to Production-Scale Infrastructure

Join Cloud Native New York City for a special double-feature evening focused on the two most critical stages of the modern software lifecycle: rapid development integration and safe production scaling.

We have expanded our lineup to bring you two specialized sessions that tackle the "Day 1" hurdles of AI-assisted development and the "Day 2" complexities of platform migrations at scale.

Session 1: Safety as a Deliverable: A Practical Pattern for Karpenter Rollouts

Speaker: Abhinav Dubey

This talk shares a practical approach to validating and rolling out Karpenter by treating safety as a deliverable. Abhinav will detail a strategy centered on a staged proving ground, gating signals, and the "hand-picked batch" pattern designed to catch problems early.

Key Takeaways:

• A rollout pattern for high-blast-radius platform changes.

• A decision framework for go/no-go under uncertainty.

• A minimal safety telemetry set that makes the rollback routine.

Session 2: Bottlenecks All the Way Down: Why AI Code Gets Stuck in your CI/CD Pipeline

Speaker: Adna Zujo

AI has drastically lowered the cost of writing code, but in modern distributed systems, integration remains the hardest, slowest, and most expensive step in shipping software. Just because AI agents can write code faster doesn't mean they don't have to wait for every step of a CI/CD pipeline to complete.

Key Takeaways:

• A practical blueprint for collaborating with AI agents in complex environments.

• How to bypass shared staging queues using mirrord.

• Strategies for shipping reliable, high-quality code at AI speed.

Evening Agenda

• 6:00 PM: Doors Open

• 6:30 PM: Session 1: Safety as a Deliverable: A Practical Pattern for Karpenter Rollouts + Q&A

• 7:30 PM: Session 2: Bottlenecks All the Way Down: Why AI Code Gets Stuck in your CI/CD Pipeline + Q&A

• 8:15 PM: Networking & Closing Remarks