Main Purpose
UbiOps is a platform designed for data scientists and teams to easily deploy, train, and run their own machine learning (ML) and data science code, as well as deploy off-the-shelf models and run data processing tasks.
Key Features
- Deployment: Deploy ML models, functions, and scripts as microservices with their own API endpoints.
- Pipelines: Build workflows based on deployments to create data pipelines.
- Requests: Trigger single runs of deployments or pipelines using data payloads.
- Experiment: Define training setups, including environment, instance type, and storage options.
- Training Runs: Execute multiple training jobs in parallel within an experiment.
- Environments: Manage code and environment separately, reuse environments for different deployments and experiments, and customize environments with additional dependencies.
- WebApp: Access UbiOps through an easy-to-use web application.
- API: Use the platform API directly to interact with UbiOps.
- Python and R Client Libraries: Integrate UbiOps API methods in your code and automate workflows using Python or R client libraries.
- Command Line Interface: Interact with the UbiOps platform API from the terminal.
Use Case
- Deploying and serving machine learning models.
- Running data processing tasks.
- Building data pipelines.
- Training and experimenting with machine learning models.
