01 / Project
Atlas
A data-driven platform for exploring complex urban systems.
2026 · Software Engineer · Web Application
Overview
Atlas is a research and planning tool for cities that need to read density, infrastructure, and movement as a single picture. It brings together spatial datasets, civic records, and modelled projections into one calm interface — designed for planners who work between maps, tables, and long-form reports.
The Problem
Urban teams were working across disconnected GIS tools, spreadsheets, and slide decks. Important relationships — between housing, transit, and public space — were visible only after hours of manual assembly. The work was slow, and the resulting documents were difficult to interrogate.
The Approach
The product was designed as a layered atlas rather than a dashboard. Each view begins with geography, then reveals quantitative context without covering the map. Filters are treated as editorial controls: few, precise, and always reversible. The system favours comparison over spectacle — two years, two districts, one question.
Implementation
A Next.js application reads curated spatial layers from PostgreSQL with PostGIS. Vector tiles are served through MapLibre for responsive cartography. Heavy aggregations run as versioned queries rather than client-side computation. TypeScript models keep layer schemas explicit so new datasets can be introduced without rewriting the interface.
Outcome
Planning groups can move from question to comparative view in a single sitting. The platform reduced the time required to assemble a district brief, and made the underlying data available as a shared reference rather than a one-off export.
Technology
Next.js · TypeScript · PostgreSQL · MapLibre
