<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Devkumar</title>
    <description>The latest articles on DEV Community by Devkumar (@devkumar_5694077be5de605a).</description>
    <link>https://dev.to/devkumar_5694077be5de605a</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4226128%2F47e86cf0-73bd-4e49-badc-54e0bccb87b7.png</url>
      <title>DEV Community: Devkumar</title>
      <link>https://dev.to/devkumar_5694077be5de605a</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/devkumar_5694077be5de605a"/>
    <language>en</language>
    <item>
      <title>Four layers for testing an LLM feature in CI: golden set, code checks, judge, gate</title>
      <dc:creator>Devkumar</dc:creator>
      <pubDate>Fri, 13 Nov 2026 10:45:39 +0000</pubDate>
      <link>https://dev.to/devkumar_5694077be5de605a/four-layers-for-testing-an-llm-feature-in-ci-golden-set-code-checks-judge-gate-20kp</link>
      <guid>https://dev.to/devkumar_5694077be5de605a/four-layers-for-testing-an-llm-feature-in-ci-golden-set-code-checks-judge-gate-20kp</guid>
      <description>&lt;p&gt;Writing&lt;/p&gt;

&lt;p&gt;Originally published on the TestDino blog. I work there.&lt;/p&gt;

&lt;p&gt;Testing an LLM feature requires more than checking whether the output looks correct. A reliable evaluation process needs four layers:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Golden dataset: Use real examples to establish expected outputs.&lt;/li&gt;
&lt;li&gt;Deterministic checks: Validate things like required fields, word counts, and specific content.&lt;/li&gt;
&lt;li&gt;LLM judge: Evaluate subjective quality using clear scoring criteria.&lt;/li&gt;
&lt;li&gt;CI gate: Run evaluations automatically and catch regressions before release.
This approach combines promptfoo evaluations with Playwright tests and GitHub Actions.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The key takeaway: a test that fails once and passes on retry is still useful data. Don't automatically dismiss it as noise.&lt;/p&gt;

&lt;p&gt;Read the complete guide for the configuration examples, evaluation workflow, and practical implementation details.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://testdino.com/blog/llm-evaluation-framework" rel="noopener noreferrer"&gt;https://testdino.com/blog/llm-evaluation-framework&lt;/a&gt;&lt;/p&gt;

</description>
      <category>llm</category>
    </item>
    <item>
      <title>TestDino articles for Software Testing Notes</title>
      <dc:creator>Devkumar</dc:creator>
      <pubDate>Wed, 04 Nov 2026 12:05:16 +0000</pubDate>
      <link>https://dev.to/devkumar_5694077be5de605a/testdino-articles-for-software-testing-notes-3a58</link>
      <guid>https://dev.to/devkumar_5694077be5de605a/testdino-articles-for-software-testing-notes-3a58</guid>
      <description>&lt;p&gt;I enjoy following Software Testing Notes and wanted to share two recent articles from the TestDino team that might interest your readers.&lt;/p&gt;

&lt;p&gt;Shift Left vs Shift Right Testing: &lt;a href="https://testdino.com/blog/shift-left-vs-shift-right-testing" rel="noopener noreferrer"&gt;https://testdino.com/blog/shift-left-vs-shift-right-testing&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Manual Testing vs Automation Testing: &lt;a href="https://testdino.com/blog/manual-testing-vs-automation-testing" rel="noopener noreferrer"&gt;https://testdino.com/blog/manual-testing-vs-automation-testing&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Both cover practical QA workflows and decision-making approaches.&lt;/p&gt;

&lt;p&gt;Would you consider including either in an upcoming issue?&lt;/p&gt;

&lt;p&gt;Thanks!&lt;/p&gt;

</description>
    </item>
  </channel>
</rss>
