Worked on modularml/mojo and modular/modular repositories to enhance performance benchmarking and telemetry systems using Python. Refactored benchmark metrics by introducing a BaseBenchmarkResult, replacing nested data models to streamline result handling and improve maintainability. This approach enabled consistent cross-task performance analytics and reduced technical debt. In the modular repository, delivered telemetry enhancements for the Performance Dashboard by adding prefill and decode statistics, persisting key performance signals to the database, and integrating with Prometheus for actionable insights. Leveraged backend development, data modeling, and data visualization skills to improve observability, support data-driven decision-making, and lay the foundation for future scalability and onboarding.