Back to Projects
Full-Stack ApplicationActive Development

BorderPulse

Real-time US-Mexico border crossing wait times, historical patterns, and predictions.

Next.jsTypeScriptPythonFastAPIPostgreSQLTimescaleDBRedisDocker

Overview

BorderPulse polls the CBP Border Wait Times API every 5 minutes, normalizes and deduplicates the data, and stores observations in a TimescaleDB hypertable. The API serves live wait times from a Redis cache, and the frontend auto-refreshes every 60 seconds. Predictions use day-of-week × hour-of-day medians computed from the last 90 days of observations. It targets people who cross the border regularly: commuters, families, and truckers.

The Problem

Border crossing wait times are unpredictable. Commuters and families waste hours without knowing which port or time is optimal.

The Solution

A full-stack application that ingests real-time CBP data, stores historical observations, and generates predictions based on 90 days of patterns, covering 23 ports across CA, AZ, and TX.

The Result

Covers 23 border ports with live updates every 60 seconds. Predictions become useful after about a week of data collection.

Key Features

  • Real-time data pipeline polling CBP API every 5 minutes
  • TimescaleDB hypertable for efficient time-series storage
  • Redis caching layer with PostgreSQL fallback
  • Prediction engine using 90-day historical medians
  • Covers 23 US-Mexico border ports across 3 states

My Role

Solo Developer

Status

Active Development

Tech Stack

Next.jsTypeScriptPythonFastAPIPostgreSQLTimescaleDBRedisDocker