Notes from the home lab
Field notes on an always-on personal AI assistant: local context, hybrid inference, dependable automations, and the homeserver work that keeps it running. Plans are labelled as plans; measurements include their setup.
A practical architecture for an always-on personal AI assistant: keep memory, permissions, triggers, and automations at home while routing inference to the right local or cloud model.
Read the article →The plan for turning a UGREEN server into a useful base for local LLMs, automations, and self-hosted tools—without pretending the hard parts disappear.
Read the article →A practical framework for choosing home automations, constraining the model, and making every run observable and recoverable.
Read the article →A simple decision framework for privacy, latency, model capability, reliability, and the real cost of operating inference at home.
Read the article →Turn vague repository advice into specific, scoped instructions with observable outcomes. Includes a worked example and a practical review checklist.
Read the article →Build a small evaluation set from real repository work: clear requirements, reproducible starting points, independent checks, and no leaked solutions.
Read the article →