Worked on the confident-ai/deepeval repository to enhance telemetry reliability by implementing the Telemetry Global Provider Integrity Guard. Focused on ensuring that the test-run tracer provider could not override the global OpenTelemetry provider, this work preserved data fidelity and reduced the risk of telemetry regressions during continuous integration. Leveraged Python and OpenTelemetry, applying test-driven development and robust version-control practices to deliver targeted bug fixes and improved test coverage for telemetry-related components. The approach emphasized focused test design to maintain observability confidence, resulting in more reliable telemetry data and a stronger foundation for future development and monitoring within the project.