<?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>webgpu on AI to Z: Navigating the Alphabet Soup</title>
    <link>https://kshreyas.dev/tags/webgpu/</link>
    <description>Recent content in webgpu on AI to Z: Navigating the Alphabet Soup</description>
    <image>
      <title>AI to Z: Navigating the Alphabet Soup</title>
      <url>https://kshreyas.dev/apple-touch-icon.png</url>
      <link>https://kshreyas.dev/apple-touch-icon.png</link>
    </image>
    <generator>Hugo -- gohugo.io</generator>
    <language>en</language>
    <lastBuildDate>Tue, 06 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://kshreyas.dev/tags/webgpu/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Check the Fridge Before You Wake Gemma</title>
      <link>https://kshreyas.dev/post/system-one-two/</link>
      <pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate>
      
      <guid>https://kshreyas.dev/post/system-one-two/</guid>
      <description>EmbeddingGemma 2 finds your memories, a small decision model decides whether one already answers you, and Gemma 4 only runs when it has to. All in the browser.</description>
    </item>
    
    <item>
      <title>बोल — Shipping Marathi TTS in Your Browser</title>
      <link>https://kshreyas.dev/post/bol-tts-marathi/</link>
      <pubDate>Tue, 05 May 2026 00:00:00 +0000</pubDate>
      
      <guid>https://kshreyas.dev/post/bol-tts-marathi/</guid>
      <description>I shipped a 326 MB Marathi text-to-speech model that runs entirely in your browser via WebGPU — no server, no API keys, just &amp;lt;audio&amp;gt; and a phoneme tokenizer in TypeScript. Type Marathi (or Marathi mixed with English — Minglish), pick a voice, hit synthesize.
Try it yourself: huggingface.co/spaces/shreyask/bol-tts-marathi. Requires WebGPU (Chrome 113+, Edge 113+). The model is ~326 MB, loads once, then cached in your browser. First load takes ~30s; subsequent visits are instant.</description>
    </item>
    
    <item>
      <title>I Gave an AI a Vocabulary of Only Emoji — And It Started to Communicate</title>
      <link>https://kshreyas.dev/post/pantheon-ui/</link>
      <pubDate>Sat, 11 Apr 2026 00:00:00 +0000</pubDate>
      
      <guid>https://kshreyas.dev/post/pantheon-ui/</guid>
      <description>In AMC&amp;rsquo;s Pantheon, a scientist named David Kim is uploaded — his mind digitized, his body gone. His daughter discovers he&amp;rsquo;s still there, trapped inside a computer, when he starts sending her emoji through her phone. He can think in full sentences. He can remember. He can love. But the only channel he has to reach her is the emoji keyboard.
I couldn&amp;rsquo;t stop thinking about that scene. So I built it.</description>
    </item>
    
  </channel>
</rss>
