FAST
SIMPLE
MODERN
Company

Build the fastest geospatial software.
Make it the easiest to use.

That’s the mission. Rapid Geospatial Systems was founded in Irvine, California, in 2026 to rethink every layer of geospatial software — from the engine underneath to the experience on top.

01Relentless optimization

Performance is never finished.

We benchmark, profile, instrument, and tune continuously. Every bottleneck is an invitation to go faster, and every improvement becomes the new baseline.

02Free to rethink

Start with what’s possible now.

We choose the algorithms, data structures, formats, and abstractions that make sense for today’s hardware, data, and users — not the ones history handed us.

03Built end to end

We don’t outsource performance.

From high-performance I/O and memory to our expression engine, data utilities, spatial indexes, and spatial algorithms, rgSpatial is built in-house in Rust. Owning the performance-critical layers lets us eliminate dependency bottlenecks and optimize the system as a whole.

04Prove it first

Claims are cheap. Evidence isn’t.

We verify correctness, document the methodology, and make sure the result can be reproduced before we put a number on the page.

Why now

Computing changed. GIS engines kept their old foundations.

Many of the spatial engines used today were designed before the modern computing stack existed. Then the pieces changed.

1990s–2000s
Legacy spatial engines object-oriented geometry · row-oriented data · single-threaded execution
The pieces arrived
2015
Rustmodern systems programming
2016
Apache Arrowcolumnar memory
2017
WebAssemblyportable native-speed compute
2023
GeoArrowcolumnar geometry
Then we put them together
2026
rg
Built after all of them.
Pure Rust I/O + spatial engine · Arrow-native · parallel by design
One engine across browser · Python · R · server

Built in Irvine. Built from scratch.

Rapid Geospatial Systems is a small engineering team that believed the industry’s tools deserved rebuilding — so we rebuilt them: the I/O, the memory model, the spatial kernels, the expression engine, even the date utilities.

/// Compute area for each geometry. Native path for Polygon/MultiPolygon
/// avoids WKB parsing; other types return 0 or delegate to WKB.
pub fn area(&self) -> Vec<f64> {
    match self {
        GeoArray::Wkb(a) => {
            par_map_range(a.len(), |i| {
                if a.is_null(i) { f64::NAN }
                else { crate::wkb::area_one(a.value(i)) }
            })
        }
        …
Not a wrapper. Not a fork. A new engine. Every kernel on this site — the parsers, the geometry, the joins, the renderer’s feed — was written here.
2026
Founded
Irvine, CA
Headquarters
One engine
Everywhere
Browser · Python · R · Server

Reach us at [email protected].

See what we’ve built.

Request a Demo