Developed real-time tokens-per-second metrics for chat statistics in the DearVa/Everywhere repository, enhancing both the UI and backend to display live performance data as token counts changed. Centralized timing logic within the ChatService improved maintainability and reduced code duplication, while careful management of event subscriptions ensured proper disposal and minimized memory leaks. Leveraging C#, XAML, and software architecture skills, the work delivered a memory-conscious design that improved system stability. These updates provided operators with enhanced observability and actionable performance metrics, supporting data-driven capacity planning and reducing maintenance overhead by streamlining telemetry and performance monitoring across the application’s chat activity.