Quidne

Azure applications, release automation and AI systems, delivered with the evidence that they work.

See the work Start a conversation

Philosophy

Build the smallest thing that can be proven to work, then grow it on evidence. Before we write code we agree, in writing:

  • What the system has to do, stated as something checkable
  • The evidence that says it does it
  • What it costs to run, measured rather than assumed
  • What happens when it fails, decided before it does
  • Who runs it after we leave, and what they need from us to do that

We hold ourselves to it here first: this site ships hand-written HTML, CSS and JavaScript, minified at release and never bundled; its analytics are cookieless and collect no personal data; its only other third-party request is the bot check on the contact form, made only when you reach it. No release is finished until an automated gate proves the deployed bytes are the ones we built.

46 checks run against every deploy of this site as of

Services

We build and run cloud applications for small and mid-sized organizations, as a senior engineer embedded with your team or as the whole team for a scoped delivery.

  • Cloud application engineering
    • Web applications, APIs and background jobs your team can still change a year from now
    • The right Azure hosting for the workload — Static Web Apps, Functions or Container Apps — chosen on cost and operational fit
    • Front ends built on web standards, so there is no framework to migrate off later
  • DevOps and release engineering
    • Every environment reproducible from one Bicep definition, so DEV and PROD cannot drift apart
    • Pipelines that build once and promote that same artifact, so what you tested is what ships
    • Automated verification gates, so a green deploy is evidence rather than a claim
    • Domain, DNS and TLS migrations with no downtime, the way this site moved
  • AI solution architecture
    • Retrieval and agent workflows scoped to one business problem, so you can tell whether they paid off
    • Evaluation harnesses, so a model change is a measured difference rather than a hunch
    • Cost and latency budgets set before a system ships, not after

Contact

Have a system that needs to work, not just ship? Tell us about it.

Quidne (Latin, why not?), independent since 2000. Anacortes, WA.

LinkedIn