I have a new most coveted possession: an NVIDIA RTX 4080 signed by Jensen Huang; the grand prize of Vercel+NVIDIA's World's Shortest Hackathon.
The winner of the @nvidiaai × @vercel world's shortest hackathon #2…
— Guillermo Rauch (@rauchg) January 18, 2025
A NYC live webcam directory with vector search, built in 2 hours with @v0 and @supabase 🔥 pic.twitter.com/4qgrvYcftX
The Hack
In two hours, we built NYCerebro, a vector-search powered look through the lens of New York City's public traffic cams. The pitch is simple: we like X-Men and have always wanted mutant powers... what would be cooler than channeling Charles Xavier's ability to look through the eyes of other mutants with his Cerebro machine?
How it Works
The "magic" is that we use OpenAI's CLIP model to embed a semantic representation of each traffic camera's current image and then compare that with a the text vector of the user's search query. By indexing all of the camera images' vectors in a vector database we can find the "most similar" images for a search query.
time squareThe rest is just some sugar. We had v0 use some basic JavaScript to show the camera (after proxying it through the server to get around CORS issues) and Mapbox to display a heatmap. Then we asked it to make the site look "cyberpunk" and used Apple's Image Playground to generate the background image.
How we Built It
The craziest part of this hack? All of the frontend code was 100% written with Vercel's v0! With some detailed prompting, a bit of back and forth debugging, and the occasional escalation to OpenAI o1 (and copy/pasting back its reply to v0) we were able to create a novel app in just two hours.
For the backend, we used a Roboflow Workflow to calculate the CLIP embeddings and a Custom Python Block to save the results to our Supabase database (with pgvector as the vector store).
Next Steps
After the hackathon we cleaned things up a bit to add new features like making the heatmap points clickable and caching the search vectors for quicker loading. We open sourced the prompts and v0's output (give it a star if you liked this post).
And now we're going to build a rig around the Jensen-signed 4080 to run some AI demos in Roboflow's SF office.
Cite this Post
Use the following entry to cite this post in your research:
Brad Dwyer. (Jan 22, 2025). NYCerebro: Winning a GPU Signed by Jensen Huang. Roboflow Blog: https://blog.roboflow.com/nycerebro/
Discuss this Post
If you have any questions about this blog post, start a discussion on the Roboflow Forum.