Worked on enhancing observability for LangChain integration within the Arize-ai/openinference repository by implementing propagation of LLM tool call attributes from individual runs to their parent agent or chain spans. This approach improved end-to-end traceability and enabled faster debugging and more transparent SLA monitoring for LangChain-based inference pipelines. Leveraging Python for backend development, the work focused on distributed tracing techniques to ensure that tool call metadata was accurately reflected in tracing user interfaces. The solution facilitated more reliable performance monitoring and root-cause analysis, demonstrating a strong understanding of observability patterns and commit-driven delivery in modern inference system architectures.