Bayou Alert
AI-Powered Flood Response Platform for Houston
Real-time dashboard for emergency responders—aggregating critical flood data, predicting high-risk areas, and delivering a responsive, interactive map UI. Built during Houston’s 2025 flood crisis to save lives and modernize disaster response.

The Problem: Houston Floods in Crisis
Houston faces some of the worst flood risk in the US. In extreme weather,
distress signals come from 911 calls, radio, social media, and government sensors—all siloed,
scattered, and hard to synthesize.
During the historic 2025 Houston flood, rescue teams were
overwhelmed and dispatch was delayed. A lack of real-time, unified information put both citizens
and first responders at risk.
- 35k+ Rescue requests in 48 hours
- 17 Bayous at major flood stage
- 1 system to bring it all together

My Role
Tech Lead & Project Architect — Led end-to-end solution delivery: system architecture, backend, data engineering, frontend dashboard, and DevOps. Coordinated with Code for Houston, set project vision, and drove implementation.
Project Goals
- Unify distress signals from sensors, social feeds, and citizens
- Predict flood surges before they become emergencies
- Empower responders to prioritize high-risk zones fast

Solution Overview
Bayou Alert brings together live sensor ingestion, a fast serverless backend, and a modern React dashboard. This enables first responders to track and predict flood emergencies with unprecedented speed and clarity.
Backend (Azure & Python)
- Live USGS sensor data ingestion (TimerTrigger)
- Cosmos DB alert storage
- API for live and historical alerts
- NLP parsing for urgent SMS (in progress)
Frontend (React + Mapbox)
- Responsive dashboard: map, tables, and charts
- Color-coded markers by severity
- Dark/light mode toggle for dispatchers
- Citizen & sensor alert aggregation

Tech Stack & Tools








Solution Overview

- Scheduled USGS flood sensor ingestion
- Cosmos DB stores real-time flood alerts
- Endpoints for alert retrieval and citizen flood reports
- NLP pipeline (planned): urgent SMS/text parsing

- Responsive React dashboard (Mapbox, Table, Charts)
- Dark/light toggle for emergency center use
- Live map view of flood sensor data & alerts
- Alert list and chart view for rapid scanning
Implementation Highlights
Live Sensor Ingestion
Automated flood level readings from Buffalo Bayou’s USGS sensors, processed in real time via Azure Functions.
Mapbox Visualization
Custom React component renders a Mapbox map with color-coded flood markers by severity—instantly scannable for first responders.
Cosmos DB Storage
Alerts and sensor data are structured, timestamped, and geolocated in Cosmos DB for fast queries and history.
Dark/Light Emergency UI
Responsive dashboard with quick dark/light toggle for usability under any dispatch center lighting.
React Architecture
Modular, maintainable code using React hooks, utility CSS, and clean separation of components for rapid iterations.
Deployment, Reflection & Impact
Deployment & Hosting
- Frontend deployed on Vercel (React + Vite)
- Backend live via Azure Functions
- Database: Cosmos DB (NoSQL)
- GitHub Repository: [Link Coming Soon]
- Live Demo: [Demo Link Coming Soon]
Reflection & Impact
Bayou Alert isn’t just a demo—it’s a real crisis tool designed with first responders, for first responders. The process challenged my skills across cloud, data, UX, and civic collaboration. From mapping sensors in Mapbox to building real-time dashboards, I learned how AI, thoughtful design, and rapid prototyping can make public safety smarter and faster.
- Developed during a real Houston flood event
- Partnered with civic tech groups & Code for Houston
- Bridged data silos (911, sensors, social) in one interface
- Optimized for urgent use in high-stress environments