Writing
Posts.
Notes on agentic AI, test automation, homelab infrastructure, and lessons learned along the way.
-
The testing pyramid grew up
How the test automation pyramid has shifted over the years, why shift-left testing matters, and what catching bugs early buys you in practice.
#testing#shift-left#engineering#ci-cd -
Evals are the new unit tests
Why asserting equals doesn't work for LLMs, how LLM-as-judge fills the gap, and where evals plug into CI/CD as the quality gate for AI features.
#evals#llm-as-judge#ai#testing -
Not all databases are the same
Why I stopped reaching for Postgres for everything, what ClickHouse is built for, and how it fits into an OpenTelemetry observability stack.
#clickhouse#postgres#databases#observability -
Agentic AI Fundamentals: From Models to Agents
I tried to structure it so each concept builds on the last, starting from what models actually are, through transformers and context windows, then into agents, tools, and MCP. Basically the stuff I wish someone had connected for me when I was getting up to speed on how things like Claude Code actually work under the hood.
#agentic-ai#claude-code#video#education -
Challenging my mental model about agentic AI
How Claude Code changed my approach to learning and weekend projects, from building an open-source agentic AI template to reconstructing my homelab for long-horizon agents.
#agentic-ai#claude-code#homelab#proxmox -
Setting Up a New Mac for Agentic AI Development
My go-to apps, CLI tools, and configurations for setting up a new Mac for software development and agentic AI work. From Homebrew to Claude Code, here's everything I install on day one.
#macos#developer-tools#setup#homebrew -
Whats a homelab?
Learn how to build your own homelab with Proxmox, Docker, and self-hosted services. From breaking my family's first computer to running local AI models and automation infrastructure.
#homelab#self-hosting#proxmox#docker