Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

You enabled mTLS in your service mesh. The security team is satisfied-on paper. But how do you actually know it is working? When an auditor asks for proof that mTLS is enforced across namespaces, how do you provide the evidence? In this demo, we focus on the verification and troubleshooting side of mTLS that rarely gets enough attention. Using Istio Ambient Mesh and Calico's Unified Platform, we look at some common misconfigurations and how to fix them..

Most Kubernetes security efforts focus on what comes in-ingress rules, admission controllers, network policies for east-west traffic. But the back door is often left wide open. By default, every pod in your cluster can reach any IP address on the internet, and in many environments, that default is never changed. This is exactly the gap attackers exploit: a compromised workload with unrestricted egress can exfiltrate data, establish command-and-control channels, or pivot to external systems without triggering a single alert.

Every organization can get a model running in a notebook. The hard part is everything that comes after: promoting that model to a production service that is secure, observable, resilient, and governed. The gap between "it works on my laptop" and "it runs in production" is where most AI initiatives stall-not because of model quality, but because the platform is not ready.

You've successfully deployed Istio Ambient Mesh to secure pod-to-pod traffic with zero-trust mTLS-but now you need to layer in sophisticated Layer 7 controls. In this demo we show how you get L7 traffic management without reverting to sidecar injection. You will see how Waypoint Proxies are deployed per-namespace or per-service to handle advanced routing, while the ztunnel continues to manage mTLS at the node level. The result is a clean separation of concerns.

Every organization can get a model running in a notebook. The hard part is everything that comes after: promoting that model to a production service that is secure, observable, resilient, and governed. The gap between "it works on my laptop" and "it runs in production" is where most AI initiatives stall-not because of model quality, but because the platform is not ready.