Things learned while building platforms, connecting systems,
and following an idea a little too far.
51 articles
NEWEST FIRSTClaude Code mods let you change the CLI itself: a row above the prompt, a docked sidebar, a slash command, a guard on tool calls. One of my mods shows what I am spending. The other decides which Kubernetes clusters Claude is allowed to touch and colors them red or green, and the same idea works for databases, logs and any other environment where dev and prod sit side by side.
I hooked TypeSafe's Jev up to a runner game, found an open 4B model that answers the same way, and spent the next day and night training my own version of it. Teacher data on rented GPUs, LoRA rounds on a 4080 Super, a public release on Hugging Face and GitHub, and a submission to JevBench.
Part 6 of the self-hosting series: connecting Ollama chat to GrowthBook through a Docker MCP server, with flag evaluation, Cloudflare Tunnel routing, and tool calls in Pages Functions.
Moving a platform onto new infrastructure without changing the hostname everyone already uses. I built the two-origin version on my own domain first, with Cloudflare Load Balancing, two Docker containers and a pair of tunnels, so the cutover is a config change at the edge that reverses in seconds.
Rhythm Hands is a little browser rhythm game I built for my kid. This post wires a physical drum to it: a piezo sensor on an ESP32 DevKit, a threshold-and-debounce sketch, and Web Serial reading hit events straight into the game with no server in between. Includes the two bugs that ate the afternoon.
Part 5 of the self-hosting series: a round AMOLED ESP32-S3 on my desk that records my voice, transcribes it, asks my local Qwen, and speaks the answer back. The full pin map, the two audio chips nobody explains, and the clock-register byte that cost me a day of silence.
Part 4 of the self-hosting series: wiring Redis's agent-memory-server into the site's AI chats, entirely on local hardware. Durable memory with no login, an exam coach that remembers your weak spots, and the mistakes I made getting there.
Part 3 of the local-LLM saga: when a friend's GPU goes offline, I stood up my own Qwen on Ollama, exposed it over a Cloudflare Tunnel locked behind Access, and wired a GrowthBook flag so the blog can flip between his box and mine from the chat header.
How this blog's AI features route through a Cloudflare AI Gateway to a friend's locally hosted Qwen model behind LiteLLM, with Turnstile and HMAC session tokens keeping bots off the GPU.
OpenCode is an open source AI coding agent with provider flexibility, cross-repo context, and per-session cost tracking. Here's how to set it up and why GitHub's pricing changes make it worth a look.
GitHub Agentic Workflows let you describe repo tasks in plain Markdown and have a coding agent run them on a schedule. Here's how they work and what you can do with them.
How I built a Claude Code skill that turns my IDE into a multi-agent development studio for building, debugging, and shipping features on my AI DevOps assistant.
How I turned a single chatbot into a team of specialized AI agents that route requests, hand off between each other, and investigate Kubernetes issues through structured workflows.
Building an AI assistant with Claude that can actually manage Kubernetes clusters - checking status, starting and stopping AKS, and handling Crossplane webhooks automatically.
A walkthrough of building a specialized AI assistant for Kubernetes, Docker, and cloud questions using Claude API, React, and Cloudflare Pages Functions with PAT authentication.
Why separate promotion workflows are unnecessary: building production infrastructure into your golden path template from the start with optional toggles.
Home Last year, I decided to rebuild my personal blog from scratch. What started as “I’ll just make a simple Gatsby site” turned into something I can only describe as a digital…
Step-by-step guide to implementing Auth0 OAuth authentication in Backstage, replacing insecure guest authentication with proper identity management for both local development and production Kubernetes deployments.
Automatic ephemeral environments for every pull request: Istio multi-cluster routing, cross-cluster communication, and GitOps-driven preview deployments.
Building a working golden path from concept to production: shared development clusters, namespace isolation, and end-to-end GitOps deployment with Backstage.
Examining the three-tier abstraction model, progressive disclosure, and environment lifecycle patterns that appear across platform engineering implementations.
A practical guide to implementing a hub-spoke Kubernetes architecture with Istio service mesh, separating control plane services from application workloads for better fault isolation and scalability.
AI coding assistants are becoming autonomous agents that modify code, run commands, and make architectural decisions. But there's no standard way to tell them what they shouldn't do. Here's a proposal.
A step-by-step guide to configuring ArgoCD for private GitHub repositories and private container images from GHCR.
Golden paths get attention for service creation. The harder problem is day-2 operations—where reality shows up.
How we made container logs visible in Kibana using Elastic Fleet Agent, why some setups silently fail, and what it takes to fix them.
A deep dive into building a self-service portal that lets developers provision Azure resources through blueprints, while Terraform and GitOps handle everything behind the scenes.
Every pull request gets its own isolated environment with HTTPS, automatic cleanup, and zero manual intervention.
Neon platforming, SMW-style scoring, sub-levels, and controller support — all running in your browser.
Introducing Terminal Hacker, a retro-inspired browser game hidden in this blog. Learn how it works, why it's fun, and what challenges await across Seasons 1–5.
How I used a Raspberry Pi, K3s, and Cloudflare Tunnel to securely publish the Kubernetes Dashboard without port forwarding.
How the Memphis Meshview project integrates Meshtastic MQTT messages into a live map and persistent chat logs, giving the community real-time insight into the mesh.
How I built a 3-node personal mesh network using Meshtastic, including a mobile, solar-powered, and desk node — plus lessons learned about node roles, placement, and settings.
A deep dive into how I built this Gatsby-powered blog, Dockerized it, pushed it to my Raspberry Pi container registry, and deployed it securely via Helm on K3s.
By Chris House, Technical Architect Spinning up a .NET application locally in a corporate environment often turns into an exercise in frustration. Developers deal with inconsistent SDK…
How Dev Containers solve the 'it works on my machine' problem: standardizing .NET development with Docker, VS Code, and secure corporate setup (NuGet feeds, HTTPS certs, Zscaler).
A deep, command-heavy walkthrough to export the MicroK8s Observability stack, split the manifests, and rebuild them as a reusable, parameterized Helm chart with sensible values, helpers, and migration tips.
A deep dive into migrating from Docker Desktop to Podman Desktop, what I learned along the way, and how experimenting with Podman Quadlets eventually pushed me toward Kubernetes for orchestration.
Step-by-step guide to migrate a Node.js project from npm to Bun for faster builds and dependency management.
How to set up the Azure Kubernetes Service MCP server for Claude Code.
How to create Agent Skills for VS Code Copilot and Claude Code.
Feature flags (also known as feature toggles or feature switches) are a software development technique that turns certain functionality on and off during runtime, without deploying new…
My goal is to use Nx to help the overhead of managing 15+ angular applications that all make up one product. This means one package.json to deal with. I have came across N x.dev’s…
How we configured Azure Pipelines to run Nx-optimized CI and trigger application-specific CD pipelines for a large Angular monorepo.
Wouldn't it be nice to notify your users when a new application build has been deployed? The first step in doing so, is to stamp your build version from the build pipeline into an asset…
Katalon is a powerful test automation solution that provides a way to test your app end to end. Azure DevOps provides developer services for support teams to plan work, collaborate on…
Firebase Cloud Firestore is “the next generation of the Realtime Database with more powerful queries and automatic scaling.” It is was acquired by Google in 2014, and is a realtime…
Transclusion is the inclusion of part or all of a component into one or more other components via a template reference. Transclusion is usually performed when the referencing component…
Sometimes in Angular templates, we may be tempted to bind a function in the html template: https://medium.com/media/f2b6de96b9bb6e3a757893ea1aee34b6/href getColor() calls a function to…
We can create prettier imports by defining the paths and baseUrl properties in the compilerOptions section in thetsconfig.json file. Given the below code, we could refactor into one…
A complete, example-driven explanation of Angular's forRoot() convention: why it exists, how it prevents multiple service instances across lazy-loaded modules, and best practices.