AWS Cloud9
Cloud IDE to write, run, and debug code in the browser
Open-Source Projects
Discover top open-source software, updated regularly with real-world adoption signals.
Discover top open-source software, updated regularly with real-world adoption signals.
Instantly spin up reproducible AI/ML container environments
envd is a CLI tool that creates isolated, OCI‑compatible container environments for AI/ML projects, supporting local, Kubernetes, and remote builds with caching and reusable Python‑based manifests.
envd streamlines the creation of AI/ML development environments by letting developers declare dependencies in a simple Python‑style manifest. With a single envd up command, it builds an OCI‑compatible image that isolates all required tools—Python, Conda, CUDA, shells, and more—eliminating the fragile, manual Dockerfile process.
Targeted at data scientists, ML engineers, and research teams, envd works on local machines, Kubernetes clusters, or remote build servers. It leverages BuildKit for fast, cached builds, reusing previously downloaded packages from PyPI or APT caches. Images can be pushed to any container registry, enabling easy sharing across teams.
Whether you need a quick local sandbox, a scalable cluster‑based training job, or a remote build on a high‑performance server, envd provides a consistent workflow. The tool also supports importing reusable functions from Git repositories, fostering knowledge reuse within organizations.
When teams consider envd, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Prototype a new model locally
Spin up an isolated container with required Python packages and CUDA in minutes, ensuring the prototype runs identically on any machine.
Scale training on a Kubernetes cluster
Build the environment once, push the OCI image, and run distributed training jobs without re‑configuring dependencies.
Share reproducible research
Publish the envd image to a registry, allowing collaborators to pull and reproduce experiments with a single command.
Offload heavy builds to a cloud server
Use remote build mode to compile the environment on a powerful server, then retrieve the cached image for local development.
envd works on any system with Docker 20.10+ and supports local execution, Kubernetes clusters, and remote build servers.
No. envd uses a Python‑style manifest to declare packages and configuration, handling Dockerfile generation internally.
BuildKit caches pip, conda, and apt downloads, so repeated builds reuse existing layers instead of re‑downloading packages.
Yes. The resulting OCI image can be pushed to any container registry (Docker Hub, Harbor, etc.) for team-wide reuse.
While manifests are written in Python, envd also supports R, Julia, and other languages via appropriate install functions.
Project at a glance
ActiveLast synced 4 days ago
