Worked on the DataDog/integrations-core repository to enhance GPU monitoring by improving metric documentation and metadata management. Focused on clarifying metric definitions, adding missing units, and marking deprecated metrics in favor of updated replacements to align with Agent 7.77 changes. Used Python and YAML to update documentation and metadata, ensuring that GPU metrics such as sm_active, gr_engine_active, and nvlink.speed are accurately described and discoverable. Collaborated with the team to coordinate changes, reducing ambiguity for both users and automated systems. This work improved monitoring accuracy and enabled more effective dashboards, alerts, and data-driven decisions for GPU performance analysis.