Descriptions:
Sally Ann O’Malley, a senior engineer at Red Hat with a decade of experience in containers, Linux security, and Kubernetes, delivers a conference talk at AI Engineer on running OpenClaw — an open-source AI agent framework — in containers, scaling from a local Podman/Docker setup all the way to OpenShift and Kubernetes.
O’Malley walks through the practical case for containerizing AI workloads: reproducibility, secret isolation, portability across x86 and Apple Silicon, and clean backup/recovery via named volumes. She uses her personal “forever claw” configuration — which includes sub-agents for Jyotish astrology readings and Boston Bruins daily briefings — as a concrete test bed for container and MCP server configuration patterns. A real-world data point from Nvidia appears mid-talk: a colleague is running OpenClaw across a 10-engineer Kubernetes cluster for model evaluations, reportedly compressing the output of six engineers into one.
The session also sketches an enterprise deployment vision: a curated, company-approved OpenClaw base image pre-loaded with approved MCP servers, team-specific skills, and sanctioned authentication credentials — a clean baseline new hires can pull and run immediately. For practitioners already comfortable with containers who want a practical on-ramp to deploying AI agents at scale, this talk covers installer patterns, volume mounting for agent skill directories, and the security framing Red Hat is building around OpenClaw on OpenShift.
📺 Source: AI Engineer · Published May 22, 2026
🏷️ Format: Tutorial Demo
