<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Harikrishna Sahu - Writing</title>
    <link>https://harikrishnasahu.com/writing/</link>
    <description>Materials informatics engineer working at the intersection of computational materials science, polymer informatics, and machine learning.</description>
    <language>en</language>
    <atom:link href="https://harikrishnasahu.com/feed.xml" rel="self" type="application/rss+xml" />
    
    <lastBuildDate>Sat, 01 Aug 2026 16:00:00 GMT</lastBuildDate>
    
    
    <item>
      <title>What T5 Adds Beyond Decoder-Only Polymer Models</title>
      <link>https://harikrishnasahu.com/writing/2026-08-01-what-t5-adds-to-polymer-language-models/</link>
      <guid isPermaLink="true">https://harikrishnasahu.com/writing/2026-08-01-what-t5-adds-to-polymer-language-models/</guid>
      <pubDate>Sat, 01 Aug 2026 16:00:00 GMT</pubDate>
      <description>A technical but accessible note on why encoder-decoder T5 architectures matter for polymer property prediction, conditional generation, molecule-text translation, and materials discovery workflows.</description>
      
      <category>POLYT5</category>
      
      <category>T5</category>
      
      <category>polymer-language-models</category>
      
      <category>generative-design</category>
      
      <category>materials-informatics</category>
      
    </item>
    
    <item>
      <title>Hardness Prediction Needs the Load</title>
      <link>https://harikrishnasahu.com/writing/2026-07-17-load-dependent-hardness-prediction/</link>
      <guid isPermaLink="true">https://harikrishnasahu.com/writing/2026-07-17-load-dependent-hardness-prediction/</guid>
      <pubDate>Sat, 18 Jul 2026 03:30:00 GMT</pubDate>
      <description>A note on our Applied Physics A paper showing why load-dependent experimental data can matter more than DFT-derived hardness proxies for Vickers hardness prediction.</description>
      
      <category>hardness</category>
      
      <category>machine-learning</category>
      
      <category>gaussian-process-regression</category>
      
      <category>materials-informatics</category>
      
    </item>
    
    <item>
      <title>Why I Built matgpr for Materials Informatics</title>
      <link>https://harikrishnasahu.com/writing/2026-07-05-why-matgpr/</link>
      <guid isPermaLink="true">https://harikrishnasahu.com/writing/2026-07-05-why-matgpr/</guid>
      <pubDate>Sun, 05 Jul 2026 17:00:00 GMT</pubDate>
      <description>A first note on matgpr, an open-source toolkit for uncertainty-aware, physics-informed Gaussian Process Regression workflows in materials informatics.</description>
      
      <category>matgpr</category>
      
      <category>gaussian-process-regression</category>
      
      <category>physics-informed-ml</category>
      
      <category>materials-informatics</category>
      
    </item>
    
    <item>
      <title>POLYT5: Turning Polymer Design Into a Text-to-Text Problem</title>
      <link>https://harikrishnasahu.com/writing/2026-07-02-polyt5-generative-polymer-design/</link>
      <guid isPermaLink="true">https://harikrishnasahu.com/writing/2026-07-02-polyt5-generative-polymer-design/</guid>
      <pubDate>Thu, 02 Jul 2026 22:00:00 GMT</pubDate>
      <description>A first note on POLYT5, a polymer-native encoder-decoder language model for property prediction, conditional generation, and high-temperature dielectric polymer design.</description>
      
      <category>POLYT5</category>
      
      <category>polymer-design</category>
      
      <category>language-models</category>
      
      <category>materials-informatics</category>
      
    </item>
    
  </channel>
</rss>
