Pipeline & Prompts

Ran Ollama in a Podman container on a Mac, hit a VM memory limit, and compared real container memory usage against Entry 01's bare-metal numbers.

How Pipeline & Prompts handles your data.

Connected a local MCP server exposing semantic search over a Chroma index to Claude Code and confirmed the tool actually gets invoked end-to-end.

An IPsec VPN throughput wall at under 1 Gbps turned out to be single-threaded encryption and ECMP flow hashing — not bandwidth.

Replaced truncation with real chunking in a local RAG pipeline — fixed content loss, but introduced a retrieval bias toward documents with more chunks.

Ran real and off-topic questions against a local Chroma vector store to see if semantic search actually discriminates relevant from irrelevant content.

Built a Python script that embeds local markdown files and a live web article into a local vector store using Ollama and Chroma.

A 45-minute payments outage and a stalled node drain reveal what managed Kubernetes actually changes — and what it doesn't — moving to ROSA HCP.

Ran llama3.2:1b at three quantization levels on Ollama and compared memory footprint and answer quality.

Why our MCP-based OpenShift agent shipped read-only, and what a failed remediation test taught us about stale-knowledge risk.

Ran llama3.2:1b and llama3.2:3b on Ollama and compared real memory footprint across model sizes.

Most hybrid cloud architectures are accidents.

Built an Ollama Modelfile that restricts a local LLM to answering only with real oc CLI commands, then tested where it held and where it broke.

Ran llama3.2:1b on Ollama and measured real memory footprint vs download size.

A NOT NULL column with no default broke a zero-downtime OpenShift deploy in 45 seconds.

A hardcoded deprecated API in a Helm sub-chart passed CI and broke OpenShift production.

Automate OpenShift incident triage with n8n: parallel RAG and MCP lookups, Vertex AI analysis, and execution limits that stop runaway LLM agents.

LLMs generate plausible averages, not environment-specific configs.

ROSA, ARO, and OSD are production-grade managed OpenShift.

Prompts are unversioned infrastructure.

How a missing pull secret delayed a major platform release — and what it taught a team about RBAC, rotation lag, and who owns the 2 AM approval.

A flaky 503 shouldn't page anyone at 2am.

AI agents that guess at your cluster state from training data will eventually be wrong.

Build a RAG pipeline in Python using FastAPI, OpenAI, and Chroma to make your internal runbooks semantically searchable.

GitHub Actions orchestrates when a deployment happens; HAProxy controls how traffic shifts to it.

Five AI touch points for OpenShift and Kubernetes: what's working today, what's still hype, and the governance layer required before any AI reaches production.

How Cloud, DevOps, and AI converge to create AIOps - automated infrastructure and intelligent systems that anticipate business needs.

Learn how AI is evolving from answering questions to taking autonomous action with goal-driven agents that execute multi-step workflows.

AI explained in plain English.

Replace manual cloud configuration with automated, version-controlled scripts.

Bridging the gap between shipping containers and software pods: See why Kubernetes is the backbone of the next-generation digital supply chain.

Learn how CI/CD pipelines automate the journey from developer laptop to production with real-world examples anyone can understand.

Docker and containers explained for beginners.

Git explained for beginners.

Linux basics for beginners.

DevOps explained in plain English.