trelay - Railway Journey Analysis Platform

Architecture Overview

Frontend - React 19 Backend - Python & FastAPI Database - SQLite (caching) AI Integration - OpenAI GPT

Background:

I was motivated to build this tool after experiencing frequent train delays on the way to work. Initially as part of a hackathon and now as a developed project I wanted a way to search a particular route and generate histograms of delays for departure and arrival times.

This project integrates with the UK Historical Service Performance (HSP) API to fetch delay data
across any route, then visualizes departure and arrival performance through interactive histograms. The AI component (OpenAI) provides travel recommendations based on the statistical
patterns - which is currently deactivated for cost reasons.

Data is cached for re-calling specific journeys but otherwise ephemerel and limited to a few GB. Once the backend is called there is station code autocomplete. This converts suggested names into the required station codes which HSP works on. The histogram breakdown shows on-time rates, early arrivals, and various delay buckets to help commuters make informed decisions.

View Source Code