vault / meta / index.md

Start here

Meta 39 words 6 outgoing 0 backlinks

These notes are short companions to the posts, not a separate set of claims. Start with a concept, follow its related notes, then open the article for the example and the limits. For code and worked projects, see Projects.

LINKS IN THIS NOTE

[[ Agents ]] In the assistant described in my posts, a model chooses steps while retrieval supplies the evidence. That creates several ways to fail: a poor tool choice, missing documents, or an answer that goes beyond them. A fluent reply alone does... [[ A/B testing ]] Random assignment creates comparable groups; a shared metric defines what is compared. A p-value describes how surprising the data would be under a specified null hypothesis. It is not the probability that the hypothesis is true. Decide the measurement and... [[ Model serving ]] A trained model is not yet a service. A shared serving path handles the repeated work around it: authentication, tracing, deployment conventions and orchestration. Keep the supported path narrow and explain failures clearly. The post distinguishes the interface I built... [[ Kubernetes ]] Kubernetes stores desired state while controllers keep trying to make the running system match it. A Deployment manages replicas, a Service gives traffic a stable destination, and Helm packages configuration. That mental model is more useful than memorising object names... [[ FM synthesis ]] In FM synthesis, one oscillator modulates another to change its spectrum. The connection to explore-exploit in the post is an analogy, not the same algorithm: a bandit learns from rewards, while an oscillator does not evaluate whether a sound is... [[ Exposure triangle ]] Aperture, shutter speed and ISO change exposure with different costs: depth of field, motion blur and noise. The post compares that tradeoff with learning rate, batch size and warmup. It is a teaching analogy, not a formula that maps camera...