Blog Posts

Latest articles and insights

A practical design for testing queue-backed counters from message publication to the user interface.

E2E Queue Count Testing: Proving Eventual Consistency Without Flaky Tests

Răzvan Mareș8 min read

A queue count looks like the smallest possible product feature: place work on a queue, show the number somewhere in the application, and verify that the number changes. In an end-to-end test, it beco...

PostgreSQLci-cd.NET+2
Design end-to-end tests that validate real user outcomes without becoming a brittle maintenance burden.

The Second E2E Topic: Turning Browser Tests Into Release Evidence

Răzvan Mareș9 min read

End-to-end testing is often introduced as the final safety net: open a browser, click through an application, and confirm that the important path works. That description is accurate, but incomplete. ...

GitHub Actionsci-cd.NET+2
Use architecture, tests, tooling, and documentation to turn a codebase into a better AI collaborator.

Improve Claude by Making Your Repository Easier to Read

Răzvan Mareș8 min read

Claude does not need a more elaborate prompt for every task. Often, it needs a repository that explains itself. That distinction matters. Developers frequently treat AI output quality as a conversati...

GitHub Actionsci-cdTypeScript+2
A practical guide to modeling finite domains with compiler-enforced alternatives.

Beyond Enums: C# Unions and Closed Classes in .NET 11 Preview

Răzvan Mareș8 min read

Most C# domain models begin with classes, inheritance, and enums. That combination is familiar, flexible, and often good enough. It also leaves an important gap: many business concepts are not merely ...

.NETef-corecsharp+1
Treat model output as a probabilistic subsystem, not a smarter function call.

AI Inference Architecture: Building Software Around Uncertainty

Răzvan Mareș8 min read

Most software systems are built around a comforting assumption: given the same valid input, a component should produce the same correct output. Advanced AI features break that assumption. A language...

PostgreSQLci-cdcsharp+2
Turn long-running AI coding sessions into recoverable, reviewable engineering work.

Claude Code Session Design: Checkpoints, Handoffs, and Safe Long Tasks

Răzvan Mareș9 min read

Claude Code is most effective when treated less like an autocomplete tool and more like a temporary engineering environment. It can inspect a repository, trace behavior across layers, propose a plan, ...

ci-cdnodejsTypeScript+2