<?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: Ansh Kansagra</title>
    <description>The latest articles on DEV Community by Ansh Kansagra (@ansh_kansagra).</description>
    <link>https://dev.to/ansh_kansagra</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%2F4051950%2F84f88c22-3c31-4989-8f11-faa12ea004ec.jpg</url>
      <title>DEV Community: Ansh Kansagra</title>
      <link>https://dev.to/ansh_kansagra</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/ansh_kansagra"/>
    <language>en</language>
    <item>
      <title># I got tired of hunting for free ML/robotics papers across 5 sites, so I built an aggregator</title>
      <dc:creator>Ansh Kansagra</dc:creator>
      <pubDate>Tue, 28 Jul 2026 19:12:38 +0000</pubDate>
      <link>https://dev.to/ansh_kansagra/-i-got-tired-of-hunting-for-free-mlrobotics-papers-across-5-sites-so-i-built-an-aggregator-1104</link>
      <guid>https://dev.to/ansh_kansagra/-i-got-tired-of-hunting-for-free-mlrobotics-papers-across-5-sites-so-i-built-an-aggregator-1104</guid>
      <description>&lt;p&gt;I'm a final-year Electronics &amp;amp; Communication Engineering student about to start a&lt;br&gt;
robotics-focused master's. For the last two years, "finding a paper" has meant&lt;br&gt;
checking arXiv, then IEEE, then MDPI, then Google Scholar, then still not being sure&lt;br&gt;
if the PDF link I found was actually free or just a paywall in disguise.&lt;/p&gt;

&lt;p&gt;So I built &lt;strong&gt;Cortexa&lt;/strong&gt; — a free, open-access-only search engine for robotics, ML,&lt;br&gt;
deep learning, computer vision, VLSI, and adjacent research. No login wall to search,&lt;br&gt;
no scraped paywalled content pretending to be free.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Stack:&lt;/strong&gt; Next.js 16 (App Router), TypeScript, Tailwind CSS v4, Supabase (Postgres +&lt;br&gt;
Auth + Storage), deployed on Vercel with daily cron ingestion.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Where the papers come from:&lt;/strong&gt; arXiv's Atom API, CrossRef (MDPI, individually&lt;br&gt;
open-access IEEE Transactions articles, and other OA publishers), and OpenAlex —&lt;br&gt;
about 22,800 papers indexed as I write this, growing daily.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The bug that actually taught me the most:&lt;/strong&gt; early on, some "View free PDF" links&lt;br&gt;
were opening raw XML instead of a PDF. Turned out CrossRef and OpenAlex sometimes&lt;br&gt;
list a publisher's &lt;em&gt;metadata API endpoint&lt;/em&gt; as the "PDF" link, not the paper itself.&lt;br&gt;
I ended up writing a small validator that rejects anything shaped like an API call&lt;br&gt;
(&lt;code&gt;api.*&lt;/code&gt; hosts, &lt;code&gt;/content/article/&lt;/code&gt; paths, &lt;code&gt;httpAccept&lt;/code&gt; query params) before it's&lt;br&gt;
ever shown as a download link — a good reminder that "the API said it's a PDF"&lt;br&gt;
and "it's actually a PDF" are different claims.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Licensing was the part I didn't expect to spend so much time on.&lt;/strong&gt; "Open access"&lt;br&gt;
isn't one thing — MDPI is 100% OA by policy, but IEEE and most other publishers only&lt;br&gt;
have it OA when the &lt;em&gt;author&lt;/em&gt; specifically paid for it. So every non-MDPI paper only&lt;br&gt;
gets ingested if its own CrossRef record carries a verified Creative Commons license,&lt;br&gt;
regardless of which journal it's technically in. That's how the index includes&lt;br&gt;
individually-open-access articles from otherwise-paywalled IEEE Transactions journals&lt;br&gt;
without misrepresenting the paywalled majority of that same journal as free.&lt;/p&gt;

&lt;p&gt;It's live at &lt;strong&gt;automata-index.vercel.app&lt;/strong&gt; (yes, the rename to Cortexa is still in&lt;br&gt;
progress — long story). Free, no ads yet, built solo. If you're in robotics/ML/AI&lt;br&gt;
research and want to kick the tires or tell me what's missing, I'd genuinely like&lt;br&gt;
to hear it.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>python</category>
      <category>claude</category>
      <category>beginners</category>
    </item>
  </channel>
</rss>
