envd is a machine learning environment tool designed to facilitate the development, training, and serving of machine learning models within Docker containers. With the ever-increasing complexity of machine learning projects, reproducibility, isolation, and ease of deployment/serving have become paramount concerns for data scientists and developers alike. envd addresses these challenges by creating a standardized and isolated environment within Docker containers, offering a seamless integration of machine learning frameworks, libraries, and the popular Ops infrastructures. With the simplified Starlark front-end, users can launch a container with Python-like syntax configuration files that can be tracked by version control tools. envd uses Buildkit as its back-end to take full advantage of the state-of-the-art artifact builder features. This presentation aims to provide a comprehensive overview of the capabilities and benefits of envd, demonstrating how it simplifies and streamlines the machine learning workflow for developers.