Ember Ready

Inspiration

Wildfire and earthquake risk info exists, but it's scattered across government tools built for scientists, not regular people. Every app I found either buried you in raw data or gave you a scary number with no explanation. I wanted something honest, calm, and actually useful.

What it does

Wildfire risk: type any US address, get a score from USDA burn-probability data plus live wind/humidity/dry-spell conditions and NWS alerts. Earthquake risk: USGS's national seismic hazard map layered with live earthquake markers, filterable by day/week/month. A personalized plan: before/during/after checklists tailored to your household. Optional AI explanations: Gemini restates the app's own numbers in plain language, never adds new facts.

How we built it

Python/FastAPI backend, plain HTML/CSS/JS frontend, SQLite caching, Leaflet maps, deployed on Vercel. Built with Claude Code from requirements I wrote and directed. Every data source is real and named and was researched and under went a credebility check that was monitored by myself.

Challenges we ran into

• Seismic data silently failed to load. An HTTP client wasn't following a redirect, hidden by a broad error handler.
• Earthquake map's coastal buffer scaled with map size instead of a fixed distance, so it barely filtered anything.
• Switching Plan tabs reset a completed checklist's progress bar to empty, even though the saved data was still correct.

Accomplishment that we're proud of

• Two full hazard tools running on real, live government data with zero fabricated numbers.

What we learned

Government APIs are messier than they look, redirects and silent failures need real testing to catch. Using AI well means writing precise requirements and testing hard, not just accepting the output.

What's next for Ember Ready

Finer-resolution hazard data instead of county-scale. Expanding the AI feature into a Q&A about your own plan, still limited to explaining, never inventing.

Built With

• claude
• css
• fastapi
• gemini
• html
• javascript
• leaflet.js
• noaa/national
• open-meteo
• python
• sqlite
• usda
• usgs
• vercel