Docker Image

A Docker image is an immutable template used to create Docker containers. It packages an application together with all of its dependencies, configuration files, environment variables, and the operating system filesystem into a single artifact that can be reliably deployed across different environments.

Docker images are built in layers, where each layer represents a set of filesystem changes defined by instructions in a Dockerfile. This layered architecture enables efficient storage and transfer, since layers shared between images are only stored once. Images are distributed through registries like Docker Hub, making it straightforward to share and deploy applications.

In AI and ML workflows, Docker images are essential for reproducibility and deployment. They ensure that a model trained in one environment runs identically in another, eliminating "works on my machine" problems. Common use cases include packaging inference servers, bundling models with their serving frameworks, creating consistent training environments, and deploying ML pipelines to cloud platforms like AWS, GCP, or Azure.