| title: Riprap NYC (Personal Mirror, L4) | |
| emoji: 🌊 | |
| colorFrom: blue | |
| colorTo: indigo | |
| sdk: docker | |
| pinned: false | |
| short_description: NYC flood-exposure briefings on L4 (self-contained). | |
| # Riprap — NYC flood-exposure briefings (L4 self-contained mirror) | |
| This Space is a self-contained mirror of | |
| [`github.com/msradam/riprap-nyc`](https://github.com/msradam/riprap-nyc). | |
| It runs on a single L4 GPU and co-hosts everything in one container: | |
| Granite 4.1 8B (via Ollama), Prithvi-EO 2.0 NYC-Pluvial, TerraMind | |
| LULC + Buildings LoRAs, and Granite TTM r2 — no external droplet | |
| dependency. Sleeps on idle; first request after sleep takes ~45–60 s | |
| to wake. | |
| The hackathon submission Space (CPU UI, droplet proxy) lives at | |
| [`AMD-hackathon/riprap-nyc`](https://lablab-ai-amd-developer-hackathon-riprap-nyc.hf.space). | |
| Apache 2.0. See the GitHub repo for full source, architecture | |
| deep-dive, methodology, and licence map. | |