Innovative approaches to challenges are at the core of what we do, and we like to spend some of our time developing experimental tools and projects to solve these challenges.

In this edition of Groundwork we’re sharing some of our latest projects that we've worked on with partners and the open-source community. Lately we've focused on surfacing Earth data in easier and more intuitive ways - with interactive viewers for massive datasets and smarter search tools for STAC. A number of projects have now been deployed, while others are laying the groundwork for the next wave of cloud-native projects.

Here’s a snapshot of what we’ve been up to:

2. Spatial Access Measures Lab

A client-side stack for fast spatial analysis.

We published an interactive app that makes Statistics Canada Spatial Access Measures on healthcare, employment, and education more accessible. The lab is a concrete use case of transforming non-cloud-optimized archives. We used DuckDB-Wasm, Deck.GL, GeoArrow, and GeoParquet — to serve this data without a server entirely in the browser. It’s a blueprint for how to use some of the more recent developments in cloud-native geospatial.

Explore the app.

3. STAC Semantic Search

Ask questions. Get STAC results.

We built a natural language search interface for STAC Items using LLM embeddings and simple vector search. A demo is live with STAC Browser integration coming soon. This opens the door to more human-accessible search experiences — from “snowpack in April” to “Sentinel-2 cloudless scenes in Ethiopia.”

• Explore the Streamlit demo
• Our STAC browser fork experiment (we want to upstream it!)

5. Dynamic STAC without the database, powered by GeoParquet

Fast, local queries over static STAC metadata.

Pete's been experimenting with how to use GeoParquet to enable fast search of STAC metadata — no Postgres required. We built a FastAPI-based prototype that can be bundled with static files and queried instantly, ideal for edge deployment or portable datasets.

This effort drove the creation of rustac, a new Rust library for working with STAC Items — which powers parsing and filtering in the API backend. It’s designed for reuse in high-performance or embedded contexts and may soon support even broader tooling.

• Check out the repo
• Read our write-up
• Powered by rustac

python -m pip install 'stac-fastapi-geoparquet[serve]'
export STAC_FASTAPI_GEOPARQUET_HREF=path/to/naip.parquet
uvicorn stac_fastapi.geoparquet.main:app

Join the Fun

From building a playground for spatial filters in Rust, to rendering Zarr files in the browser, to experimenting with STAC metadata without a database — these projects are all riffs on the same theme: making geospatial data more usable, performant, and human-friendly.

None of these are finished products. They’re experiments born out of real needs, weird ideas, and collaborations. If something here sparks your interest — let us know. Better yet, fork a repo, open an issue, or try one out in the wild. Get in touch if you have an idea or a sticky problem. We're excited to see where these go next.