Worked on enhancing observability for AI completion interactions in the 0xPlaygrounds/rig repository by integrating OpenTelemetry tracing into the Rust-based backend. Focused on API and backend development, the approach involved implementing trait-based response handling and token usage tracking to capture detailed telemetry data across various provider paths. Developed new traits to standardize tracing and response metadata collection, enabling consistent performance metrics and facilitating data-driven optimization. Added unit tests to validate trait implementations and ensured code consistency through formatting. The work prioritized reliability and measurable performance gains, supporting proactive monitoring and faster diagnosis of AI workflows without addressing major bug fixes.