I'm a final-year Electronics & Communication Engineering student about to start a
robotics-focused master's. For the last two years, "finding a paper" has meant
checking arXiv, then IEEE, then MDPI, then Google Scholar, then still not being sure
if the PDF link I found was actually free or just a paywall in disguise.
So I built Cortexa — a free, open-access-only search engine for robotics, ML,
deep learning, computer vision, VLSI, and adjacent research. No login wall to search,
no scraped paywalled content pretending to be free.
Stack: Next.js 16 (App Router), TypeScript, Tailwind CSS v4, Supabase (Postgres +
Auth + Storage), deployed on Vercel with daily cron ingestion.
Where the papers come from: arXiv's Atom API, CrossRef (MDPI, individually
open-access IEEE Transactions articles, and other OA publishers), and OpenAlex —
about 22,800 papers indexed as I write this, growing daily.
The bug that actually taught me the most: early on, some "View free PDF" links
were opening raw XML instead of a PDF. Turned out CrossRef and OpenAlex sometimes
list a publisher's metadata API endpoint as the "PDF" link, not the paper itself.
I ended up writing a small validator that rejects anything shaped like an API call
(api.* hosts, /content/article/ paths, httpAccept query params) before it's
ever shown as a download link — a good reminder that "the API said it's a PDF"
and "it's actually a PDF" are different claims.
Licensing was the part I didn't expect to spend so much time on. "Open access"
isn't one thing — MDPI is 100% OA by policy, but IEEE and most other publishers only
have it OA when the author specifically paid for it. So every non-MDPI paper only
gets ingested if its own CrossRef record carries a verified Creative Commons license,
regardless of which journal it's technically in. That's how the index includes
individually-open-access articles from otherwise-paywalled IEEE Transactions journals
without misrepresenting the paywalled majority of that same journal as free.
It's live at automata-index.vercel.app (yes, the rename to Cortexa is still in
progress — long story). Free, no ads yet, built solo. If you're in robotics/ML/AI
research and want to kick the tires or tell me what's missing, I'd genuinely like
to hear it.
Top comments (0)