mikelev.in

I’m drawing a line from the basic idea of sequenced actions, like a Turing machine or even wax-on/wax-off, to my current need for deeper website insights than GitHub Pages provides. To truly understand causation and correlation in my site traffic, especially with AI bots and crawlers, I need raw log files. Therefore, I’m embarking on self-hosting using NixOS, starting with ordering a new router, viewing this technical setup itself as a manageable, sequenced process that will ultimately enable mu

mikelev.in

This article captures a passionate exploration of a unique methodology for future-proofing web architecture in the age of AI. It delves into the granular realities of web mechanics, from server logs to semantic HTML, revealing a philosophy of building self-healing, machine-understandable systems. The author envisions a “Sovereign API for Intelligence” that treats the web as a dynamic file system, where insights are derived from direct observation and iterative experimentation, rather than abstra

mikelev.in

This article captures the very essence of my philosophy for building a resilient, observable home-hosting environment. It’s a deep dive into reclaiming digital sovereignty, moving beyond abstract concepts to the concrete implementation of network security. The journey from initial setup to a robust DMZ, secured by fundamental networking principles like RFC1918, truly highlights the “Forever Machine” methodology. It demonstrates how seemingly small, technical steps accumulate to create a powerful

mikelev.in

My passion lies in crafting fully automated, observable, and self-hosted digital ecosystems. This particular blueprint details the practical steps of transforming a remote server into a broadcasting node, leveraging NixOS for reproducibility and xdotool for screen choreography. It’s about achieving complete control over one’s data stream, from collection and analysis to real-time dissemination, and laying the foundation for advanced AI integration without cloud dependencies. The ability to see a

mikelev.in

Our Jekyll site’s rendering time became unmanageable due to an inefficient O(n²) nested loop approach for calculating related posts based on keywords. Drawing inspiration from a previous optimization involving a flattened post index (flat_index), we initially tried pre-processing all keywords into lookup tables but found it overly complex in Liquid. The breakthrough came with a simplified O(n) method: we leverage the existing flat_index, process only the current post’s keywords, and iterate thro