<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Writing on Eugen Miknevic</title><link>https://eugenmik.github.io/posts/</link><description>Recent content in Writing on Eugen Miknevic</description><image><title>Eugen Miknevic</title><url>https://eugenmik.github.io/images/og-default.jpg</url><link>https://eugenmik.github.io/images/og-default.jpg</link></image><generator>Hugo</generator><language>en-us</language><lastBuildDate>Fri, 09 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://eugenmik.github.io/posts/index.xml" rel="self" type="application/rss+xml"/><item><title>Publishing a casting model: what it took to make the weights runnable</title><link>https://eugenmik.github.io/posts/publishing-a-casting-model/</link><pubDate>Fri, 09 Oct 2026 00:00:00 +0000</pubDate><guid>https://eugenmik.github.io/posts/publishing-a-casting-model/</guid><description>Releasing the solidification surrogate was not a matter of uploading a checkpoint. Checking that the published files reproduce the internal ones, and reporting the results honestly, took longer than the upload.</description></item><item><title>Human-in-the-loop AI on the foundry floor</title><link>https://eugenmik.github.io/posts/human-in-the-loop-ai-foundry-floor/</link><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><guid>https://eugenmik.github.io/posts/human-in-the-loop-ai-foundry-floor/</guid><description>Five design rules from building GussCore, a foundry ERP/MES where workers report by voice and a person confirms every AI proposal.</description></item><item><title>Validating a GPU solidification solver: from +51% to within 3.7%</title><link>https://eugenmik.github.io/posts/validating-gpu-solidification-solver/</link><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><guid>https://eugenmik.github.io/posts/validating-gpu-solidification-solver/</guid><description>How I checked the FoundryFlash solver, and how the largest error turned out to be in the sand data.</description></item><item><title>Voxels or tetrahedra: what casting AI needs from the mesh</title><link>https://eugenmik.github.io/posts/voxels-or-tetrahedra/</link><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><guid>https://eugenmik.github.io/posts/voxels-or-tetrahedra/</guid><description>A staircase boundary misstates the cooling rate where feeding decisions are made. What my own voxel prototype measured, and why a tetrahedral mesh suits a neural network.</description></item></channel></rss>