parkie-doo

@parkie-doo.sh

1130 Followers 187 Following

PhD Candidate at UC Berkeley EECS // Programming Languages • Human-Computer Interaction • Visualization • Cartography // Building alpha.cartokit.dev Software Engineer at @gristnews.bsky.social // Maps • Data • Graphics • Code

A user loads a PMTiles archive combining 8 billion building footprints from Google's Open Buildings dataset, Microsoft's GlobalMLFootprints, and OpenStreetMap into cartokit. They then zoom in on the Bay Area and University of California, Berkeley and style the map as a categorical choropleth, with building footprints colored by source.
A view from the bleachers at Opening Day for the Oakland Ballers.
A user creates a choropleth map in cartokit, downloads the map to their local computer as a .ck.json file, and re-uploads the map to cartokit after refreshing their browser. On re-upload, the user sees the map exactly as it was prior to download.
A user opens opens up the Add Layer modal in cartokit, clicking the "From Gallery" tab. They add a layer of Polygon features denoting urban areas in the United States. After the layer is successfully added, they reopen the Add Layer tab and select a new layer of Point features containing the locations of magnitude 2.5+ earthquakes in the past 30 days.
A scrollytelling graphic animates to show the history of California wildfires linked to PG&E equipment, regulatory action taken against PG&E, and the frequency of fast-trip outages since. The graphic transitions between map views showing the locations of key wildfires, a timeline view highlighting CPUC fines and fast-trip outage count and frequency, and additional map views visualizing fast-trip outages as choropleth maps by circuit line.
A user hovers over features in three different layers in a cartokit map. The tooltip automatically determines the topmost interactable layer for the mouse position and dynamically displays attributes of that layer. As the user moves closer to screen edges, the tooltip adjusts its position to ensure it never goes offscreen.
A screen recording of a user using tooltips and keybindings to quickly navigate the cartokit user interface. For example, pressing E opens the program editor while pressing G renders the map in "globe" mode.
A user interacts with an interactive map visualizing intersections of lead exposure, poverty, and race in Chicago. The user clicks on individual neighborhood boundaries to see statistics on lead exposure and socioeconomic factors. The user also searches for a water service line address, getting detailed information on the composition of the water line and whether it may contain lead.
A user creates a choropleth map of results in the US 2020 presidental election using only natural language prompts in cartokit.
A user switches between four different basemaps in cartokit: the "Dark Matter" basemap from CARTO, the "Landscape Dark" basemap from MapTiler, the "Alidade Smooth Dark" basemap from Stadia Maps, and the "Watercolor" basemap from Stamen Design.
A cartokit user creates a heatmap of all 2.5 magnitude or greater earthquakes over the last calendar month, using the USGS real-time earthquake feed.
A user changes the text "PhD Student" to "PhD Candidate" in their Bluesky profile.
A screenshot of the first page of our upcoming publication, "Fast Direct Manipulation Programming with Patch-Reconciliation Correspondence", to appear at PLDI 2025.
A scroll-based animation showing how ethylene oxide off-gases inside of a truck while being transported from a sterilizer plant to a warehouse.
A user undoes and redoes various map editing actions in cartokit using ⌘+Z and ⌘+Shift+Z.
A user switches a map's view between web mercator and globe projections in cartokit.
One sequence of a scroll-based visual interactive, tracing the outline of the Yakama Nation reservation as defined by the original cession text in 1855.
A scroll-based visual interactive highlighting the history of the dispossession of Yakama Nation land. The interactive shows various historical treaty maps and federal surveys that unjustly stripped the tribe of promised reservation land.
A user toggles between the MapLibre GL JS and Mapbox GL JS backends in the cartokit compiler.
A user interacts with a drawer menu and map controls on an interactive map of state trust lands on reservations.
A user searches an interactive map of state trust lands on reservations using both place names and latitutde-longitude coordinate pairs.
A user explores an interactive map showing the rights types of state trust lands on reservations.
A user explores an interactive map of state trust lands located within tribal nations' sovereign boundaries.
A user explores an interactive map of state trust lands located within tribal nations' sovereign boundaries.
A screenshot of the hero on the cartokit docs site.
A screenshot of alpha.cartokit.dev showing the interface for adding custom vector tile basemaps via a JSON specification. The background basemap is a variant of Stamen's toner tiles using a light blue for water features and an orange for land cover.
A map developed in cartokit showing rivers of south and east Asia visualized by their flow accumulation.
The title slide for a talk entitled, "Supporting Data Journalism through Compilers for Visual Inputs." Above the words "data journalism" are three screenshots of published graphics. Beneath the word "compilers" is a depiction of a source program being transformed to a target program. Beneath the words "visual inputs" are three examples of SVG-based data visualizations.
A photograph of the audience before my talk, “Supporting Data Journalism through Compilers for Visual Inputs.” In the foreground is a laptop in presenter mode.
The title slide for a talk entitled, "Supporting Data Journalism through Compilers for Visual Inputs." Above the words "data journalism" are three screenshots of published graphics. Beneath the word "compilers" is a depiction of a source program being transformed to a target program. Beneath the words "visual inputs" are three examples of SVG-based data visualizations.
A screenshot of the reviz Chrome extension reverse engineering a data visualization published in the New York Times. On the right, a user has uploaded a dataset to the extension; simultaneously reviz has generated an output visualization that looks like the original graphic but uses the user-uploaded data.