Geospatial Software Studio · Edge platforms, technical software
A studio building data-heavy software — a cloud GIS platform for local government, LiDAR processing applications, live telemetry and AI-driven intelligence. Agents do the typing. The engineering judgment is where the time goes.
Sta 01 · Practice
Five areas, each with production work behind it. The common thread is data that has to be right — coordinates a survey depends on, elevations an assessment rests on, telemetry an operation is planned around, sources a claim has to be traceable to.
Parcel and cadastral systems for local government — vector tiles, imagery catalogues and public records, with history preserved across every reload rather than overwritten. Built to deploy as one appliance, on-premises or in a cloud.
Terrain, surface and canopy models, classification, coverage and overlap analysis, and ASPRS accuracy reporting against surveyed checkpoints. Desktop applications where the data is too large to move, and streamed pipelines where it is not.
Scheduled ingest from public feeds into edge databases, with the derived metrics computed on write rather than on read. NOAA space weather lands in D1 on a cron and is served as a risk score, not a raw table.
Collection and summarization pipelines where a language model does the reading. Feeds in, briefings out, with the source trail kept intact so any claim can be walked back to where it came from.
Applications where the processing stays on the client because the data should not leave the machine. No upload endpoint means no upload endpoint to breach.
Sta 02 · Principal work
The largest pieces of work in the studio, each given its own sheet. A platform for parcels and LiDAR in local government, now in early alpha, and two desktop applications for the survey work that feeds it.
Sheet A
Cloud GIS platform for local government
In development · early alpha
An independent, cloud-delivered GIS platform for US county and municipal government — parcels, LiDAR point clouds and orthoimagery served from PostGIS through a Cloudflare edge. Early alpha, built by a small team in Maine: the roughly 34,000-parcel test environment now runs on local VM storage, and limited free testing opens to interested municipalities by request.
Sheet B
LiDAR processing suite for Windows
v2.9.0 · in production use
A desktop application and matching CLI that turn classified LAS/LAZ point clouds into terrain, surface, canopy and vegetation-obstruction deliverables. A ground-up interface redesign over the processing engine: a navigation rail, dependency-aware workflow selection, and a run plan that states what will happen before a multi-hour job starts.
Sheet C
LiDAR delivery inspection & reporting
v6.62 · in production use
A desktop application that scans a directory of LAS files and produces a self-contained HTML report on what is actually in the delivery — point counts, densities, coverage acreage, classification and return distributions, and where files overlap each other. Where the processing suite makes deliverables, this one inspects them.
Sta 03 · Method
The differentiator is not that a model writes the code — everyone has a model. It is that the specification, the review and the deployment path are tight enough that the output can reach production the same day.
Written before any code exists. What the thing does, what it must not do, and what done looks like. Most projects built this way fail here, not later.
Agents implement against the specification. Fast enough that a wrong direction costs hours instead of weeks — which changes what is worth attempting at all.
Every line is read. Generated code fails in specific, recognizable ways: plausible logic that is subtly wrong, unhandled edges, silent fallbacks. Knowing the failure modes is the job.
Straight to the edge. Static build to Pages, request-time work to Workers, state to D1. Nothing to provision, so deployment stops being a phase.
Sta 04 · Stack
A deliberately small set, chosen once and reused. Everything below is in production across the work above — none of it is aspirational.
Sta 05 · Record
Alongside the principal work, seven projects run in public. Four are shown here; each one is live, and each links straight to the running site rather than to a case study about it.
Drone imagery GPS viewer
Coordinate reference registry
Space weather forecast & tracking
News collection & summarization
Sta 06 · Contact
Engagement
Work starts with a written specification — what the thing does, what it must not do, and what finished looks like. That document is the deliverable before any code is, and it is what makes a same-week build possible without it being a gamble.
If you have a problem that is mostly data and partly interface, that is the shape of thing this studio is built for.
[email protected]