The archive
Essays on AI, cloud containers, and security — plus a few lessons learned the hard way.
My real-time Socket.IO game dropped from 72 connected MBA students to 2 mid-class. Short ping timeouts read congested Wi-Fi as dead connections, and every investment triggered a full leaderboard broadcast. Here are the fixes (longer timeouts, debounced broadcasts, rate limiting) and the bigger lesson: AI-written code can be correct while your testing assumptions are wrong.
AI agents need the same container infrastructure as any enterprise app: containers for portability, Kubernetes for orchestration, hybrid cloud for flexibility. The hard part isn’t building agents, it’s operating them. This piece lays out four criteria for a platform that can, and where IBM Cloud Code Engine, Red Hat OpenShift, and IBM Cloud Kubernetes Service fit.
IBM Bob, IBM’s agentic AI coding assistant, plans and writes production-ready code in your IDE. IBM Cloud Code Engine then runs it as a serverless container with no Kubernetes clusters to manage, automatic TLS, and scale-to-zero pricing. Together they’re the shortest path from AI-assisted development to production, told through a reality-TV lens.
The Knicks won on fundamentals, not star power, and a container platform should be judged the same way. On IBM Cloud, the fundamentals are managed Red Hat OpenShift, Keep Your Own Key (KYOK) encryption on FIPS 140-2 Level 4 hardware, and IBM Cloud Satellite for running containers on-prem, at the edge, or on other clouds.
Secure multi-cloud applications with a platform built for hybrid security from day one, not by stitching together each provider’s tools. The three building blocks: one control plane for policy across clouds (IBM Cloud Satellite), customer-held encryption keys (Keep Your Own Key), and built-in compliance frameworks for regulated industries.
For enterprises running containers and microservices, IBM Cloud stands out in five areas: built-in security and compliance for regulated industries (PCI, HIPAA, GDPR, NIST), fully managed Kubernetes and Red Hat OpenShift, hybrid and multicloud deployment, integrated DevOps and AI tooling, and more than 60 data centers worldwide.
Platform teams choosing a cloud for OpenShift usually ask which provider is best. The better question is who owns the upgrades: how fast new Red Hat releases become available, whether support windows match Red Hat’s lifecycle, and who troubleshoots when an upgrade breaks. The piece gives criteria for comparing IBM Cloud, AWS ROSA, Azure ARO, and Google Cloud.