<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<?xml-stylesheet type="text/xsl" href="/rss.xsl"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Recsys on Abdullah Al Mamun</title>
    <link>https://www.newabdullah.com/tags/recsys/</link>
    <description>Recent content in Recsys on Abdullah Al Mamun</description>
    <image>
      <title>Abdullah Al Mamun</title>
      <url>https://www.newabdullah.com/og-default.png</url>
      <link>https://www.newabdullah.com/og-default.png</link>
    </image>
    <generator>Hugo</generator>
    <language>en</language>
    <copyright>newabdullah 2025</copyright>
    <lastBuildDate>Sat, 19 Sep 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://www.newabdullah.com/tags/recsys/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Embeddings: Representation Layer of AI</title>
      <link>https://www.newabdullah.com/posts/embeddings-representation-layer-of-ai/</link>
      <pubDate>Sat, 19 Sep 2026 00:00:00 +0000</pubDate>
      <guid>https://www.newabdullah.com/posts/embeddings-representation-layer-of-ai/</guid>
      <description>What an embedding actually is, every way people build them from matrix factorization and word2vec through two-tower retrieval to LLM-derived vectors, how embedding tables dominate recsys parameters, cosine against dot product against Euclidean, and how vector search really works with HNSW, IVF and product quantization.</description>
    </item>
  </channel>
</rss>
