← Back to Dashboard

Weather Intel

A weather intelligence platform built for a stamped concrete business in Massachusetts.

The Problem

Concrete work depends on weather. Too cold and it won't cure. Too hot and it cracks. Rain ruins a fresh pour. Wind dries the surface wrong. Sealer needs a full day of dry conditions before and after application.

Standard weather apps say "72°F, partly cloudy." That doesn't answer the question a concrete crew needs answered at 6AM: can we work today?

The Solution

This platform scores weather conditions specifically for concrete work — green, yellow, or red — for pouring, sealer application, and post-pour curing. It checks temperature, humidity, wind, precipitation, and dew point against industry thresholds and gives a go/no-go verdict with a recommended work window.

The 48-hour forecast shows exactly which hours are safe. The sealer check looks backward (did it rain in the last 24 hours?) and forward (will it rain in the next 24?). The cure check evaluates whether the next 48 hours are safe for fresh concrete to harden.

The Data

Historical weather data comes from NOAA — the same government agency that runs the National Weather Service. Their raw files are notoriously messy: fixed-width text formats, missing values coded as -9999, temperatures stored in tenths of degrees Celsius. This platform cleans, converts, and validates that data automatically.

Live forecasts come from Open-Meteo, updated every 15 minutes via a real-time streaming pipeline. Severe weather alerts from the National Weather Service are monitored and pushed to the dashboard.

18,000+
Daily Weather Records
10
MA Weather Stations
45
Jobs Tracked
60
Automated Tests

How It's Built

NOAA Raw Data → Python ETL → PostgreSQL
                             ↓
                          dbt (transform + test)
                             ↓
Airflow (orchestrate) → BigQuery (warehouse)
                             ↓
Redis Streams → FastAPI + WebSocket → Dashboard
                             ↓
GitHub Actions CI/CD → Auto-deploy to EC2
Python PostgreSQL dbt Apache Airflow Redis Streams Google BigQuery FastAPI WebSocket GitHub Actions AWS EC2 nginx Let's Encrypt

Built By

Dylan Mota — concrete contractor and data engineer based in Massachusetts. This platform is a working business tool and a portfolio project demonstrating end-to-end data engineering.

View the code on GitHub →