<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Joe Barrow</title>
  <link href="https://jbarrow.ai/" rel="alternate" type="text/html"/>
  <link href="https://jbarrow.ai/feed.xml" rel="self" type="application/atom+xml"/>
  <id>https://jbarrow.ai/</id>
  <updated>2026-06-12T00:00:00Z</updated>
  <author>
    <name>Joe Barrow</name>
  </author>
  <entry>
    <title>Searching, Fast and Slow</title>
    <link href="https://jbarrow.ai/2026-06-12-searching-fast-and-slow/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2026-06-12-searching-fast-and-slow/</id>
    <updated>2026-06-12T00:00:00Z</updated>
    <summary>Revisiting &#34;Slow Search&#34; in the age of agentic retrieval.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Building a 1-Outlet, 4-GPU Workstation</title>
    <link href="https://jbarrow.ai/2026-05-18-gpu-workstation/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2026-05-18-gpu-workstation/</id>
    <updated>2026-05-18T00:00:00Z</updated>
    <summary>As a grad student, the one thing I desperately wanted was a GPU workstation. As a gainfully employed adult I can finally make that happen.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Agents have (Information) Needs</title>
    <link href="https://jbarrow.ai/2026-05-13-agentic-information-need/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2026-05-13-agentic-information-need/</id>
    <updated>2026-05-13T00:00:00Z</updated>
    <summary>Information retrieval is about satisfying an information need, but a query is a poor stand-in. Your agent is capable of expressing one, so you should probably use it.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>OCR&#39;s Cambrian Explosion 1 - Introduction</title>
    <link href="https://jbarrow.ai/2026-05-09-ocr-cambrian/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2026-05-09-ocr-cambrian/</id>
    <updated>2026-05-12T00:00:00Z</updated>
    <summary>Strong, open VLMs enabled an explosion of open OCR model releases, with little sign of things letting up. In this survey, I detail the models, their evaluation, research trends, and open questions.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Google Gemini 102 - Advanced Structured Outputs</title>
    <link href="https://jbarrow.ai/2025-01-22-gemini-102/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2025-01-22-gemini-102/</id>
    <updated>2025-01-22T00:00:00Z</updated>
    <summary>Probing the supported output types of Gemini.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Google Gemini 101 - Object Detection</title>
    <link href="https://jbarrow.ai/2025-01-15-gemini-101/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2025-01-15-gemini-101/</id>
    <updated>2025-01-15T00:00:00Z</updated>
    <summary>Navigating Gemini&#39;s API for object detection with vision and Structured Outputs.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Be Careful Interpreting Averaged Benchmarks</title>
    <link href="https://jbarrow.ai/2025-01-09-averaged-benchmarks/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2025-01-09-averaged-benchmarks/</id>
    <updated>2025-01-09T00:00:00Z</updated>
    <summary>Thoughts on averaged benchmarks and hidden correlations.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
  <entry>
    <title>Introduction to TinyHNSW</title>
    <link href="https://jbarrow.ai/2024-12-01-introduction-to-tinyhnsw/" rel="alternate" type="text/html"/>
    <id>https://jbarrow.ai/2024-12-01-introduction-to-tinyhnsw/</id>
    <updated>2024-12-01T00:00:00Z</updated>
    <summary>The first post in the TinyHNSW series, introducing the tutorial and the library.</summary>
    <author><name>Joe Barrow</name></author>
  </entry>
</feed>