<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Onur Kirman — Writing</title><description>Field notes on agentic systems, computer vision, and production ML.</description><link>https://kirman.dev/</link><item><title>On agentic pipelines, and where they still break</title><link>https://kirman.dev/blog/agentic-pipelines-where-they-break/</link><guid isPermaLink="true">https://kirman.dev/blog/agentic-pipelines-where-they-break/</guid><description>Tool-grounded reasoning gets you most of the way. The last mile is evaluation, and nobody tells you that part up front.</description><pubDate>Sun, 14 Jun 2026 00:00:00 GMT</pubDate><category>agentic-ai</category><category>production</category></item><item><title>Diffusion models, briefly</title><link>https://kirman.dev/blog/diffusion-models-briefly/</link><guid isPermaLink="true">https://kirman.dev/blog/diffusion-models-briefly/</guid><description>A short, practical mental model for DDPMs — from a thesis I actually had to defend.</description><pubDate>Fri, 10 Apr 2026 00:00:00 GMT</pubDate><category>computer-vision</category><category>research</category></item><item><title>Notes from an OCR paper, five years later</title><link>https://kirman.dev/blog/ocr-paper-five-years-later/</link><guid isPermaLink="true">https://kirman.dev/blog/ocr-paper-five-years-later/</guid><description>What held up, what I&apos;d redo, and why segmentation is still the hard part.</description><pubDate>Sun, 01 Feb 2026 00:00:00 GMT</pubDate><category>research</category></item></channel></rss>