Nimbus is a lightweight, developer-friendly framework designed to help small teams deploy task-focused NLP models to the cloud with minimal friction. It combines the convenience of managed platforms with the affordability and control of DIY approaches.
This guide walks through a typical user journey—from installation to deployment and management of models—using the Nimbus CLI and its web-based Playground.
✅ Prerequisites
Before getting started with Nimbus, make sure your environment includes the following:
| Tool | Description | Link |
|---|---|---|
| Node.js v20+ | JavaScript runtime required to run Nimbus CLI. Includes npm. | Download Node.js |
| AWS CLI | Required to authenticate and interact with AWS. | Install AWS CLI |
| AWS CDK | Used to define and deploy cloud infrastructure as code. | AWS CDK Guide |
| Docker | Required for building and bundling model containers. | Get Docker |
ℹ️ After installing the AWS CLI, run aws configure to set up your credentials.
🚀 Getting Started
Installation
Install Nimbus globally via NPM:
npm install -g nimbusnlpℹ️ Upon first use, Nimbus will create a .nimbusStorage directory in your current working directory to store configuration and deployment artifacts. To change its location, simply navigate to your desired directory before running any CLI commands.
🔧 CLI Commands
After installation, you can access the CLI by typing:
nimbusCLIThis will display all available subcommands.
Core Commands
nimbusCLI deployDeploy a new NLP model. This also provisions the necessary infrastructure if it doesn’t exist.nimbusCLI listView all models currently deployed, along with their names, descriptions, and endpoints.nimbusCLI deleteDelete a specific model and its associated resources, both locally and in the cloud.nimbusCLI destroyTear down the entire Nimbus deployment infrastructure (API Gateway, models, storage, etc.).nimbusCLI uiLaunch the Nimbus Playground, a local web UI to view and test your deployed models.
🧠 Deploying a Model
To deploy a model, run:
nimbusCLI deployYou will be guided through a series of interactive prompts:
- Model type: pre-trained or fine-tuned
- Unique model name
- Source:
- For pre-trained: a model identifier
- For fine-tuned: a local file path
- Optional description
Upon completion:
- Nimbus will deploy the model to AWS.
- You’ll receive:
- An HTTPS endpoint URL for predictions.
- An API key to access the endpoint.
You can immediately begin sending requests to your deployed endpoint.
📋 Listing Deployed Models
To check which models are currently deployed:
nimbusCLI listYou’ll see a list of models with their associated information:
- Model name
- Description
- Endpoint URL
🌐 The Nimbus Playground
For a visual interface, run:
nimbusCLI uiThis opens the Nimbus Playground in your browser, where you can:
- View deployed models
- Select a model
- Send prediction requests directly in the UI
It’s a great tool for testing before integrating endpoints into production.
🗑️ Deleting a Model
To remove a model:
nimbusCLI deleteYou’ll be presented with a list of your deployed models. Select the one you wish to delete.
💥 Destroying All Infrastructure
To completely remove your deployment (API Gateway, Lambda functions, models, and storage):
nimbusCLI destroy📌 Final Notes
Nimbus is designed to make deploying NLP models as simple and cost-effective as possible, without sacrificing power or flexibility. Whether you’re a solo developer or a small team, Nimbus helps you go from model to API in minutes.