The easiest way to parallelize, speed up
and manage ML experiments in the cloud
Optumi is a JupyterLab extension that provisions cloud resources, executes notebooks and keeps track of past runs. No workflow changes required.
We make data scientists more productive
⚡️
Speed up data cleaning
and model training
Optumi makes offloading experiments onto powerful CPU and GPU machines effortless.
With one click we’ll dynamically provision cloud resources and replicate your environment to run on them.
Launch experiments as parallel batch jobs
🚀
Instead of waiting for models to run sequentially, you can amplify your productivity by kicking off training jobs in parallel.
You won’t need to change your workflow or deal with containers.
⚙️
Automate monotonous workflow tasks
Have you ever babysat training runs to make sure expensive machines are shut off?
Simply tell Optumi to auto-release resources and text you when jobs finish.
Keep track of past runs and spending
📒
Optumi makes prior run data easily accessible.
This includes summary information like duration and cost as well as files and notebook output.
Benefits
Faster execution ✅
Reduce processing time by leveraging powerful CPU and GPU machines.
Resources are a click away.
New experiments ✅
Freed from resource limitations,
you can work with more complex models and larger datasets.
Lower cloud bills ✅
When spinning up and down machines is effortless, there's no need to pay for them 24/7.
Library agnostic ✅
Bring your favorite Python tools and frameworks without having to understand Docker containers.
Cloud abstraction ✅
Every cloud has its own way of doing things. Leverage their innovations without learning their jargon.
Free your laptop ✅
We know there's more to your remote workday. Offloading jobs makes room for Zoom calls and Chrome tabs 🙂