An open source stack for your AI.

Core infrastructure, curated frameworks and apps, and everything in between.

What can you do with Nebari?

Nebari runs the AI and data science tools your team needs on infrastructure you control.

Run AI on infrastructure you own

Deploy into your own cloud account, or onto your own hardware. Your models, data, and prompts stay where you already keep them.

Pick a provider

Go from prototype to something your team uses

Package what you built as a Software Pack. It comes up with its own address, a login, and routing already wired in.

How packs work

Add capabilities as you need them

Install only the packs you need today. Add serving, chat, or agent workflows later, and upgrade each one on its own.

Browse the catalog

Give every teammate the same setup

Everything on the platform is declared in a git repository and reconciled from it, so the cluster your teammate opens is the one you described.

GitOps on Nebari

Watch what your AI is doing

Logs, metrics, and traces are collected from every running pack and forwarded to the observability backend of your choice.

Platform architecture

What you don't have to do

Nebari runs the layer underneath your apps, so your team doesn't have to.

Wait on someone else to deploy it

Install a pack and it comes up on its own address, behind the login your company already uses.

Software Packs

Build a login for every internal app

Colleagues sign in once with the account they already have, and you decide which groups reach which app.

Set up single sign-on

Rent a GPU box on the side

GPU machines are a node group in the same cluster, so your team uses the hardware it already pays for.

Change a running cluster

Back up your own work

Turn backups on once, and the platform copies your notebooks and files to off-cluster storage every day.

Back up and restore

Learn Kubernetes to ship your project

The platform handles what runs underneath. You work with notebooks, apps, and models.

What Nebari is

Ecosystem of tools

Pick the pieces you need. Every project is open source and designed to work on its own or together.

Nebari core

New robust core infrastructure layer for platform deployment on popular cloud providers. Security, authentication, role-based access, and observability built in by design.

Install NIC

Nebi

Tools to manage, share, and collaborate on the AI software lifecycle across your team. Includes future plans for SBoM generation and compliance checking.

Nebi on GitHub

Usage-specific UIs

Multiple UI options available: Jupyter/VSCode for developers, chat and dashboard interfaces for AI users, and more.

Software packs dashboard

AI model serving pack

Serve and monitor different AI models for your development workflow or production applications.

LLM serving pack docs

Chat assistant pack

A ready-to-use chat interface for your teams, wired into platform login and connected to the models you serve on Nebari.

Chat pack docs

Agentic AI pack

Frameworks for building, deploying, evaluating, and monitoring AI agent-based workflows.

Pi coding agent docs

Data orchestration pack

Tools for building data engineering pipelines, and governance and provenance of data and AI assets.

Coming soon

Custom integrations

All Nebari projects are developed on an open and extensible architecture that allows for quick integration of any libraries you need.

Build your own pack

Community hub

A space for community members to share their custom packs.

Join the community

Deploy anywhere

Deploy Nebari on your own infrastructure, on your local machine or the cloud of your choice. It is designed to be flexible, extensible, and vendor-agnostic. AWS, Azure, and Hetzner are supported today, with Google Cloud on the way.