HackathonNLPDockerSvelte
The pitch
Matcha matches hackathon participants to teams based on a free-text description of what they want to build and what they bring. We ran sentence embeddings on each entry and used cosine similarity to surface the best matches.
How we built it under 24 hours
We self-hosted pre-trained NLP models inside Docker so we could keep inference local — no API keys to provision, no rate limits during the demo. The Svelte frontend talked to a small FastAPI service.
Won Second Place ($1,500) at the 2023 SYNCS Hack.