Our initial reference harness is OpenClaw, but my broader work centers on the infrastructure, engineering practices, and trust mechanisms required to make agentic systems reliable on OpenShift.
- Agentic infrastructure: building Kubernetes-native capabilities for deploying, operating, and managing long-running AI agents.
- Agentic engineering: working with agents daily and developing practical expertise in execution loops, tool integration, model routing, multi-agent coordination, evaluation, and failure recovery.
- Agent memory: one of the components of agentic systems I am most interested in; exploring shared memory and context systems that allow agents to retain, retrieve, and coordinate knowledge over time.
- Trust and guardrails: designing security boundaries, observability, human oversight, and operational controls for enterprise agent deployments.
- Model serving and AI platform engineering on OpenShift with the Mass Open Cloud.
- LLM performance benchmarking, workload tracing, and observability tooling.
- Single-model, distributed, and multimodal inference using KServe, vLLM, and llm-d.
- The Open Education Project, supporting university courses running on OpenShift AI within the Mass Open Cloud.
Agentic AI · AI Infrastructure · OpenShift · Kubernetes · Agent Memory · Model Serving · Observability · Guardrails
