<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
  <channel>
    <title>Arshnoor Singh Sohi — Blog</title>
    <link>https://arshnoorsinghsohi.com/blog</link>
    <description>Writing on AI systems, data engineering, and building things.</description>
    <language>en-us</language>
    <item>
      <title>Positional Encoding in Transformers, Explained From Absolute Zero</title>
      <link>https://arshnoorsinghsohi.com/blog/transformer-positional-encoding-explained</link>
      <guid>https://arshnoorsinghsohi.com/blog/transformer-positional-encoding-explained</guid>
      <pubDate>Sun, 09 Aug 2026 00:00:00 GMT</pubDate>
      <description>Transformers read every word in a sentence at the same instant, so how do they know word order? We build positional encoding from scratch — vectors, sine waves, and a beautiful rotation trick — explaining every single term along the way, with diagrams at every step.</description>
    </item>
  </channel>
</rss>
