Worked on the foundation-model-stack/bamba repository to enhance and consolidate performance documentation for the 9B release. Focused on aligning release notes and Markdown files with observed performance metrics, the work clarified claims such as a potential 5x decode-stage speedup for long sequences and a 2.5x speedup on vLLM, while transparently outlining the conditions for these improvements. The approach emphasized clear, evidence-based communication to improve stakeholder understanding and release readiness. All contributions were documentary in nature, involving no code changes or bug fixes, and relied on strong documentation skills and proficiency with Markdown to ensure technical accuracy and clarity.