Reliable software systems, practical architecture, and AI-enabled applications.
I design and modernize software platforms, improve performance, and help teams ship reliable software without unnecessary complexity.
Selected work
Pulse checks whether a service behaves correctly from different networks, not merely whether one monitor can reach it. A central ASP.NET control plane distributes definitions to deployable agents, which return measurements over HTTP.
HTTP, ping, and TCP checks record latency and failures; the dashboard compares recent results and agent health. SQLite keeps the first deployment simple, while shared contracts keep the server, agents, and Blazor UI aligned.
Read more ...
cowork-server is a C# API for shared coworking meeting rooms. It checks access, room availability, opening hours, booking windows, duration limits, conflicts, and price snapshots before saving a reservation.
Commands and queries stay separate inside one application and database. The booking lock coordinates competing requests only within one process, so a multi-instance deployment would need another concurrency mechanism.
Read more ...
From the notebook
A timeout can leave a caller unsure whether work succeeded. Idempotency makes repeating the same operation safe for its intended business effect, even when the acknowledgement is lost.
A small C# background worker demonstrates stable keys, saved results, and duplicate handling, with explicit limits around concurrency, restarts, retention, and external effects.
engineering architecture #csharp #dotnet #idempotency #background-processing
Read more ...Follow a method that needs an HTTP response, from starting the request to returning a parsed result. See how the HTTP task and the method's own task carry different parts of the work.
The walkthrough covers what await does while the response is pending, how failures and cancellation reach the caller, and where the method continues afterward.
engineering software-development #csharp #dotnet #async
Read more ...A separate read store lets a query model match its screen or report instead of the write model. It can remove expensive joins and let read traffic scale independently.
The new boundary also creates a synchronization problem. Reads can be stale, projections must recover safely, and the system has to state what a user should expect after a command.
engineering architecture #architecture #cqrs #projections #eventual-consistency #dotnet #csharp
Read more ...Minimal CQRS keeps commands and queries in one deployed application using one database. The paths differ because commands protect state changes while queries return data shaped for a caller.
This retains transactions and immediate reads after a command. It does not provide independent read scaling, denormalized reporting projections, or isolation from a busy write workload.
engineering architecture #architecture #cqrs #cqs #dotnet #csharp
Read more ...