<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Kevin Dela Rosa — Writing</title><link>https://kevindelarosa.com/blog</link><description>Video, agents, computer vision, and personal experiments.</description><language>en</language><atom:link href="https://kevindelarosa.com/feed.xml" rel="self" type="application/rss+xml"/><item><title>I put every CSI button in a CLI</title><link>https://kevindelarosa.com/blog/i-put-every-csi-button-in-a-cli</link><guid isPermaLink="true">https://kevindelarosa.com/blog/i-put-every-csi-button-in-a-cli</guid><description>Overcast is a video OSINT agent: eyes, ears, and OSINT reach for any agent, wired to a case that remembers. My DEF CON 34 Demo Labs talk, written down.</description><pubDate>Sun, 09 Aug 2026 12:00:00 GMT</pubDate></item>
<item><title>mDex is open source: bring your own catalog</title><link>https://kevindelarosa.com/blog/mdex-personal-matcher</link><guid isPermaLink="true">https://kevindelarosa.com/blog/mdex-personal-matcher</guid><description>The offline visual matcher from the card-scanner post is now on GitHub, Apache-2.0. Point it at a folder of your own images and an iPhone recognizes those objects offline.</description><pubDate>Thu, 06 Aug 2026 12:00:00 GMT</pubDate></item>
<item><title>I vibe coded an offline card scanner for the National</title><link>https://kevindelarosa.com/blog/mdex-offline-card-scanner</link><guid isPermaLink="true">https://kevindelarosa.com/blog/mdex-offline-card-scanner</guid><description>10,023 trading cards identified and priced entirely on-device. No server, no signal, no problem. Here&apos;s the classic computer-vision rabbit hole I fell down this weekend, and how a benchmark I shipped at NeurIPS 2023 came back to stress-test the scanner.</description><pubDate>Mon, 27 Jul 2026 12:00:00 GMT</pubDate></item>
<item><title>How to grep video</title><link>https://kevindelarosa.com/blog/how-to-grep-video</link><guid isPermaLink="true">https://kevindelarosa.com/blog/how-to-grep-video</guid><description>Notes on context engineering and agent harnesses for video libraries: designing the structured representations to make media legible to LLMs.</description><pubDate>Mon, 20 Apr 2026 12:00:00 GMT</pubDate></item>
<item><title>The million-dollar slide: what your sales transcripts are missing</title><link>https://kevindelarosa.com/blog/the-million-dollar-slide</link><guid isPermaLink="true">https://kevindelarosa.com/blog/the-million-dollar-slide</guid><description>How multimodal video understanding changes what companies can extract from their video archives.</description><pubDate>Mon, 07 Jul 2025 12:00:00 GMT</pubDate></item></channel></rss>
