A simple CLI tool to deploy your Machine Learning models to cloud, with a public API to use.
Project description
Aerostat
Aerostat is a simple CLI tool to deploy your Machine Learning models to cloud, with a public API to use.
Get started
Installation
The name Aerostat
has been used by another PyPI project, please install this package with:
pip install aerostat-launcher
Once installed, it can be used directly via aerostat
. Most likely you will need to run this module with python -m
prefix since it is not included in $PATH
.
To deploy your model, there are only three commands needed: install
, login
, and deploy
.
Setup
Run the following command, and it will install all the dependencies needed to run Aerostat.
python -m aerostat install
To login to Aerostat, you need to run the following command:
python -m aerostat login
You will be prompted to choose an existing AWS credentials, or enter a new one. The AWS account used needs to have AdministratorAccess.
Deploy
To deploy your model, you need to dump your model to a file with pickle, and run the following command:
python -m aerostat deploy
You will be prompted to enter:
- the path to your model file
- the input columns of your model
- the ML library used for your model
Or you can provide these information as command line options like:
python -m aerostat deploy --model-path /path/to/model --input-columns "['col1','col2','col3']" --python-dependencies scikit-learn
Roadmap
- Deploy a model to AWS Lambda
- Improve error handling, including login checks
- Improve user interface, including rewrite prompts with Rich, use more colors and emojis
- Return deployment info and simple test demo with HTTP GET request
- Make it a pip installable package
- Handle AWS authentication from the CLI
- Support deploying to GCP
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for aerostat_launcher-0.0.5-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 949a8449a915461e5e2fdc611dc640e2732df27fad77f5f4b325ee93ba9d8991 |
|
MD5 | 1cf077f6bad33279415a298d47caa200 |
|
BLAKE2b-256 | 62e36bba7837ee5ff8c78115259ebddb0a7c598f0b894e641411396663d8a13a |