<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Software-Engineering on Adrian Spiridon</title><link>https://adrianspiridon.dev/tags/software-engineering/</link><description>Recent content in Software-Engineering on Adrian Spiridon</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 04 Sep 2026 00:00:00 +0300</lastBuildDate><atom:link href="https://adrianspiridon.dev/tags/software-engineering/index.xml" rel="self" type="application/rss+xml"/><item><title>Build-Time vs Runtime AI: An Instruction–Data Variability Framework</title><link>https://adrianspiridon.dev/blog/build-time-vs-runtime-ai/</link><pubDate>Fri, 04 Sep 2026 00:00:00 +0300</pubDate><guid>https://adrianspiridon.dev/blog/build-time-vs-runtime-ai/</guid><description>Not every feature that can use AI should keep AI in the runtime path. A useful distinction is whether the task and input space can be specified at design time. If they can, deterministic software is usually the better runtime architecture; if either remains open-ended, runtime AI starts to earn its place.</description></item><item><title>When Implementation Stops Being the Bottleneck</title><link>https://adrianspiridon.dev/blog/when-implementation-stops-being-the-bottleneck/</link><pubDate>Thu, 03 Sep 2026 00:00:00 +0300</pubDate><guid>https://adrianspiridon.dev/blog/when-implementation-stops-being-the-bottleneck/</guid><description>AI does not make every part of software delivery equally faster. As implementation accelerates, the bottleneck moves upstream to product definition and downstream to business validation — and that will push technical roles closer to the business domain.</description></item><item><title>The Generation–Evaluation Asymmetry</title><link>https://adrianspiridon.dev/blog/the-generation-evaluation-asymmetry/</link><pubDate>Wed, 02 Sep 2026 00:00:00 +0300</pubDate><guid>https://adrianspiridon.dev/blog/the-generation-evaluation-asymmetry/</guid><description>A football argument led me to a broader question: why can we often recognize a good or bad result even when we could not produce the same result ourselves? The same generation–evaluation asymmetry appears in cognition, creativity, computer science, and now AI-assisted software development.</description></item></channel></rss>