#DevOps

UPS Sizing for Homelabs: Battery Runtime, NUT Monitoring, and Safe Shutdowns (2026)

Incident Response: Writing a Postmortem People Actually Read

Learn how to conduct blameless incident postmortems, extract root causes with the 5 Whys framework, write clear timelines, and ensure action items get executed.

Zero-Downtime Blue-Green Deployments on Cloudflare Workers: Architectural Playbook

Ditch the risky big-bang releases. Learn how to implement native Gradual Deployments, version upload pipelines, and Version Affinity on Cloudflare Workers using Wrangler.

Building a Rollback Strategy for Changes an AI Agent Made Overnight

Overnight agent recovery. How to design atomic Git commit squashing, automated snapshot isolation, DB migration rollbacks, and fast revert hooks.

Building a Usage Dashboard for Your Team's AI Tool Spend

The AI FinOps telemetry dashboard tutorial. How to build real-time OpenTelemetry tracking for token usage, cost per developer, and model latency.

Building Guardrails for an Autonomous Coding Agent on a Real Repo

Engineering agent safety. How to build AST syntax boundary checks, automated test gates, Git worktree isolation, and diff size limiters in TypeScript.

Building Internal Cost Controls for AI Tool Usage Across a Team

The engineering team cost control playbook. How to build internal API proxy gateways, budget rate limiters, token usage dashboards, and model routing rules.

Building With AI Agents as Teammates: A Practical Workflow, Not Hype

A production engineering playbook. How to organize daily dev workflows around AI agents as specialized junior teammates, from spec writing to automated verification.

Cognition Bought Windsurf for $250M and Turned It Into Devin Desktop

The AI tool space has consolidated. Read a technical analysis of Cognition's acquisition of Codeium's Windsurf, the rebranding to Devin Desktop, and the evolution of agentic IDE interfaces.

Evaluating a New Model Release Before Migrating Your Production Prompts

The AI regression testing guide. How to build automated prompt eval suites, semantic similarity scoring, and LLM-as-a-Judge benchmarking in TypeScript.

Giving an AI Agent Write Access to Production: What I Learned the Hard Way

Production agent postmortem. What happened when an autonomous agent dropped a production table, and the 5 security guardrails required for production write access.

I Let Claude Code Run Autonomously for a Full Sprint. Here's What Broke

An engineering postmortem of a 2-week experiment delegating a sprint to an autonomous coding agent. Explore recursive loop failures, token burns, and execution security.