I'm a hands-on Software Architect at Southern Ute Shared Services, in Durango, Colorado, with fifteen-plus years building and owning production systems end to end: requirements and architecture through deployment, production support, and iteration. That ownership covers a portfolio of internal and external-facing products, and the work around them: buy-versus-build analysis, vendor evaluation and selection, systems integration, project management, and technical leadership. I am still in the code, not only the diagrams.
The other half of the job is AI adoption: setting the vision and strategy, running the proofs of concept, and owning the part that decides whether any of it lands, which is awareness, training, and governance. The systems and agents I build myself are on this site, live and open to be taken apart.
I came into software through environmental science and GIS rather than computer science. That is not a detour. It is where I learned to ask whether the data can carry the decision somebody is about to make on it, which is the question every system below is built around.
The route here was not computer science. Environmental studies, then remote sensing and GIS, then two county assessor's offices where parcel and land-records work turned into .NET engineering: a county web mapping application and an assessor records site that people outside the building actually used. I joined the Tribe in 2014 building the tooling landmen use to query and edit oil-and-gas leases, and running the GitHub Enterprise Server behind it. One thread runs through all of it: data integration and API work, from SOAP services early on to REST, and later the GitHub REST and GraphQL APIs. In 2018 it became architecture: Azure applications and APIs, SharePoint Online on the far side of the ECM migration, a Power Platform estate to govern rather than just build on, and the standards and reusable patterns the next project starts from. AI is the newest layer on the same stack. Each step added a layer without dropping the one underneath.
Outside work, most of my time goes into the AI side of this, and the projects on this site are the built output, not sketches. NextAurora runs live and is meant to be broken on purpose: kill a service mid-order and watch the saga finish anyway. San Juan Watershed Riparian Watch serves a public map over a satellite-data pipeline and a Random Forest versus foundation-model field test. okl is a public tool with its eval receipts committed next to the code, and Quartzose is the design record for grounded, cited answers over enterprise documents. They sit where fifteen-plus years of production software meets GIS and environmental science, and each is written up in the open, with the architecture, the tradeoffs, and the claims that didn't hold up when I re-checked them.
Recent structured study includes the Engineer's RAG Accelerator (NeoSage, 2026), agentic AI specializations from Vanderbilt and DeepLearning.AI, and the TensorFlow Developer specialization.
That is the short version. The long one is here: how the arc actually went, what the architecture, data, platform, and AI work looks like day to day, and what I believe about building systems.