Skip to content

Learn more

This page collects external resources that go deeper into how opentrash works and how to extend it. Useful if you're building something on top of the package, contributing back, or want to understand the implementation in more detail than the architecture page provides.

Course material

A free, 12-lesson course walks through building the package from scratch: GPS ingest, the routing engine, pattern detection, RouteView, and release engineering. The course is hosted by AI Research Corps (AIRC) and available at:

airesearchcorps.org/opentrash

It's an option for users who want to understand the package at the implementation level — equally suitable as a reference if you'd like to build a similar package for an adjacent domain.

Source code

The package source is on GitHub at github.com/malsheikhyass/opentrash. Issues, discussions, and pull requests are welcome.

opentrash builds on a strong open-source geospatial stack. If you're new to any of these, their documentation is the right place to go deeper:

  • DuckDB — in-process analytical database that drives every set-based aggregation in the package.
  • DuckDB Spatial extension — the spatial-join primitives the engine depends on.
  • GeoPandas — Python's standard geo-dataframe library.
  • Shapely — Python geometry library underlying GeoPandas.
  • PyProj — coordinate-reference-system transformations.
  • MapLibre GL JS — the open-source map renderer that powers RouteView output.
  • Apache Arrow / Parquet — the columnar storage format used throughout for cached and prepared data.

Background reading

Some of the design decisions in opentrash are easier to understand against the broader context of fleet telematics and geospatial analytics:

  • Geographic Information Systems for Solid Waste Management — the three-tier GIS taxonomy (operational, support, strategic) maps onto the package's static-layer / cache / product separation.
  • The DuckDB whitepapers and conference talks on columnar query execution explain why DuckDB is well-suited for the chunked CTAS pipelines used in pattern detection.

If you have additional resources to recommend, please open a pull request adding them here.