Updated 2026-08-30. Written by Christopher Gilliland, founder of ZynVoss LLC — Fort Pierce, FL.
Principles before brands
- Boring technology you can hire for
- One primary database until you must split
- Clear environments: dev / staging / prod
- Observability before microservices fashion
Common winning shape
Frontend: modern web UI (server-rendered or SPA as needed)
Backend: REST or RPC APIs with clear auth
Data: relational DB + migrations
Jobs: background workers for email/SMS/reports
Deploy: containers or PaaS with HTTPS and backups
We implement this daily: full-stack development · technologies.
FAQ
Do I need Kubernetes day one?
Almost never for early products.
Next steps
What we will actually put in production in 2026
Pick technology you can hire for in Florida. Keep one primary database until it hurts. Have dev, staging, and production that a human can name. Watch the app before you split it into twelve “microservices.” That is the whole sermon.
A shape that keeps working for small teams: a web UI (server-rendered or a modest SPA), REST APIs with boring auth, PostgreSQL or SQL Server with migrations, a worker box for email/SMS/reports, HTTPS, nightly backups, and a pipeline that is not “copy files from Chris’s desktop.” ZynVoss LLC runs IIS and ASP.NET for our own public site and portal. We use React and Node when the product is an application. We use Python for data and automation. We use LLM APIs for intake and phone agents with logs and a human path.
What we usually refuse on v1
- Kubernetes for a ten-user internal tool.
- A new database for every feature.
- Native iOS and Android before the web loop works on a phone browser.
- Five environments nobody deploys to.
Observability means: we can tell who did what, the error is in a log, and paging does not require the original author. Security means hashed passwords, secrets not in the repo, Stripe for cards, and roles enforced on the server. See technologies and full-stack.
If a vendor pitches a rewrite because your stack is “old,” ask whether they have restored a backup and whether staff can still invoice tomorrow. Fashion is not a requirement. Monday morning is.
What we run on this building
The public site and client portal are ASP.NET on IIS, on a box we control in Fort Pierce. That is not a pitch for everyone. It is what we can see, back up, and restore without a ticket to a faceless cloud queue. Azure and AWS when the client wants them. Docker when promoting staging to production would otherwise be a prayer. React when the UI is an application. Plain HTML when it is a page.
SQL Server or PostgreSQL with migrations. Secrets not in git. Passwords hashed. Stripe for cards. LLM APIs for intake and phone agents with logs and a human path — we do not dump patient notes into a prompt. Kubernetes for a ten-user tool is a hobby. See technologies (the menu) and process (how a job runs). If a vendor wants a rewrite because your stack is “old,” ask whether staff can invoice tomorrow.