<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Richard Pallangyo: Writing</title>
    <link>https://www.richardpallangyo.com/</link>
    <atom:link href="https://www.richardpallangyo.com//feed.xml" rel="self" type="application/rss+xml" />
    <description>Data and AI engineer based in Seattle. Data science, data engineering, and AI engineering, building production systems and the products around them.</description>
    <language>en-us</language>
    <lastBuildDate>Wed, 15 Apr 2026 00:00:00 GMT</lastBuildDate>
    <item>
      <title>When to abandon the low-code path</title>
      <link>https://www.richardpallangyo.com//writing/low-code-pivot</link>
      <guid isPermaLink="true">https://www.richardpallangyo.com//writing/low-code-pivot</guid>
      <pubDate>Wed, 15 Apr 2026 00:00:00 GMT</pubDate>
      <description>A pivot story from SSA Marine. I prototyped an AI agent in Microsoft Copilot Studio, hit a wall on the human-in-the-loop email approval flow, and made the costly call to rebuild from scratch with LangGraph and FastAPI. The lesson I took: always prototype the hardest workflow first, not the easiest one.</description>
      <category>AI</category>
      <category>production</category>
      <category>decisions</category>
      <category>LangGraph</category>
    </item>
    <item>
      <title>Anti-hallucination through tool grounding</title>
      <link>https://www.richardpallangyo.com//writing/grounding</link>
      <guid isPermaLink="true">https://www.richardpallangyo.com//writing/grounding</guid>
      <pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate>
      <description>When I built the conversational agent for Ubunifu Madness, my NCAA prediction platform, I had a specific problem. The model knew enough about basketball to fabricate plausible-sounding facts. The fix was structural rather than prompted. Every claim had to be backed by a tool call, and if the tool didn&apos;t exist, the agent had to say so.</description>
      <category>AI</category>
      <category>hallucination</category>
      <category>agents</category>
      <category>Claude</category>
    </item>
    <item>
      <title>Notes on AI-assisted development</title>
      <link>https://www.richardpallangyo.com//writing/claude-code</link>
      <guid isPermaLink="true">https://www.richardpallangyo.com//writing/claude-code</guid>
      <pubDate>Tue, 10 Feb 2026 00:00:00 GMT</pubDate>
      <description>A measured account of what working with AI coding tools at scale actually looks like. It covers the collaboration shape, what makes a draft good or bad, where the discipline goes, and the productivity questions I can and can&apos;t honestly answer.</description>
      <category>AI-assisted development</category>
      <category>Claude Code</category>
      <category>patterns</category>
    </item>
  </channel>
</rss>