Securing AI/ML Workflows: Optimizing Container Images in Kubernetes Environments - Wojciech Kocjan

This talk explores the critical challenges associated with building and managing container images for AI/ML workloads in cloud-native environments, particularly within Kubernetes deployments. The speaker addresses issues such as dependency management, compatibility, security risks from bloated images, and vulnerabilities posed by CVEs. By sharing insights from his experience, he focuses on optimizing container images, aiming for reduced size and zero CVEs, while overcoming testing and performance hurdles. The session covers effective techniques like minimizing unnecessary dependencies, selecting secure base images, and optimizing build times, providing a roadmap for enhancing the efficiency and security of AI/ML workflows at scale. Wojciech Kocjan, the speaker, is a Senior Software Engineer at Chainguard and a Cloud Native expert with extensive experience in public clouds and container technologies.