Worked on enhancing the Triton Inference Server ecosystem by focusing on build system management, dependency management, and release management using Python. Updated the server repository’s build configuration to incorporate the latest stable versions of ONNX Runtime and OpenVINO, improving compatibility and ensuring build stability for inference workloads. Delivered a stable release of the genai-perf package in the perf_analyzer repository, supporting reliable benchmarking for GenAI model deployments. The work emphasized maintaining up-to-date dependencies and preparing components for production readiness, with a technical approach centered on robust release processes and careful dependency upgrades to streamline deployment and benchmarking for end users.