Thoughts, insights, and updates on technology, design, and innovation.
How far can you get running AI models entirely on your own hardware? A practical look at Ollama, Apple Silicon, model selection, quantization, and building a private RAG pipeline with FastAPI and pgvector.

When you're forced to move production to a new server, most teams rush to just get things running again. Here's the phased playbook I use to make the migration itself the security upgrade not just a lift-and-shift.
Why a programmed, local-first network beats off-the-shelf smart plugs, and how to build home automation as code rather than vendor-locked app rules.
A practitioner's account of why Next.js on the front end and Python on the back end consistently deliver reliable, maintainable enterprise applications.

