Subscribe
Sign in
ZenML — Create reproducible machine learning pipelines
Ever spent hours training ML models only to have them crash and burn in production? ZenML is here to help. Track metadata, compare results, and scale to the cloud. It’s open-source, extensible, and integrates with battle-tested MLOps tooling off-the-shelf.
13
Replies
David Aronchick
Hunter
Hi all!
I'm David Aronchick, co-founder of Kubeflow, and I'm extremely honored to present ZenML. I've been chatting with this team for a while and I love the direction they're going; solving not just the training of machine learning but actually how to roll it out in production. We'd love your feedback!
Share
Hi Team,
congrats on the launch! Good sign for the German startup scene to see more deep tech!
A big thank you to our Hunter @aronchick! Hi PH community - co-creator of ZenML here. ZenML comes from a team of data scientists and engineers who struggled for years to:
a) Put models in production.
b) Keep up with the continuous onslaught of tooling in the ecosystem.
We built a system internally that lets us abstract away certain architectural + tooling choices in a pipeline-ing system based on the best tools on the market. We then spent some months trying to make it easy for the community to use and extend with custom integrations. We're launching with some useful features out-of-the-box such as cheap cloud orchestration and automated model deployment, and plan to add many more features soon.
We're looking for data scientists and ML practitioners in general to try ZenML and run their workloads on the new framework. Feedback is crucial and we look forward to improving the product based on it. Thank you and wishing you all a happy new year!
Amazing tool - have been playing around with the early versions and it improved a lot. Super handy when running many different experiments within a larger team or alone for fun! :) Keep it up :)
