The app brings together local news, location-relevant social posts, and alerts from the National Weather Service.
Engineering
Canary.ai
Local news, social updates, and weather alerts in one emergency-information app.
What I built
I led the capstone team building a React Native and Bun mobile app backed by Supabase search for live emergency-information retrieval.
The app aggregates local news, location-relevant social posts, and National Weather Service alerts, with live update notifications. We presented the project at Vanderbilt’s CS Senior Symposium.
Purpose & audience
Help people find critical, location-specific emergency information faster during severe weather events.
Built for local residents who need relevant news, social updates, weather alerts, and live notifications during emergency conditions.
Under the hood.
Follow the architecture, one connection at a time.
Canary.ai / architecture
Supabase-backed search and semantic vector embeddings support retrieval of relevant incoming information.
Relevance filtering focuses the incoming information on the user’s location rather than showing every update.
Live update notifications bring relevant emergency information into the React Native mobile experience.
Scroll to follow the connections. Select a step to take a closer look.
Decisions & tradeoffs
We integrated semantic vector embeddings to filter incoming local news, social posts, and NWS alerts by user-location relevance.
Filtering by relevance makes the volume of information more manageable. Because the sources differ, freshness and reliability need to remain visible rather than treating every update as equivalent.
Outcomes & limitations
Delivered the capstone mobile app and presented its approach at Vanderbilt’s CS Senior Symposium. The repository and symposium poster provide implementation detail.
The usefulness of the app depends on source freshness, retrieval relevance, and notification reliability. It complements the underlying sources of emergency information.
What’s next
Possible directions for further development.
Further evaluate source freshness, location relevance, and notification reliability across different emergency conditions, keeping the underlying sources visible to users.