Worked on security hardening and vulnerability remediation for Azure Machine Learning curated environments in the Azure/azureml-assets repository. Focused on upgrading OS packages and Python dependencies within Dockerfiles, pinning transitive dependencies, and introducing explicit upgrade paths to address critical CVEs. Leveraged Bash and Python to automate patching, enforce reproducible builds, and validate remediations using SBOM-driven verification. Enhanced maintainability by consolidating patching scripts, standardizing CVE remediation, and streamlining dependency management. Improved compliance and security posture by patching vendored Java libraries, upgrading PyTorch installations, and cleaning up SBOM artifacts, resulting in more secure, reproducible, and maintainable ML environment images for production workloads.