KeaML allows data scientists to create machine learning environments with no effort. You just have to select an image, the hardware specs that best suit your needs from our large catalog and click 'Create'. That's all.
Hi PH community π! I'm Gabriel, founder of KeaML.
I've been working in the AI field for the last 7 years. Along the years, I've seen some problems repeat time after time.
When I started working on data science, algorithms were that simple that any beefy laptop were more than enough for training models, but this is not the case nowadays. We can still work on our laptops, but training times would kill developers' productivity.
How did I try to solve this? Anytime I needed more computing power I went to my AWS console and started an EC2 instance. Even though I was able to train my models, I had to do a lot of heavy-lifting every time I needed to start an instance. And most importantly, everybody in the team had to do the same. Do you know what happened? Everyone ended up having environments with different libraries and libraries' versions installed, which led to making experiments reproducibility almost impossible. One day I decided to automate that manual work, and this automation then evolved to what today is KeaML.
π¨βπ» What is KeaML?
KeaML allows data scientists to create development environments with no effort. They just have to select an image (a set of libraries and dependencies) and a hardware spec. Then, the environment will be created and will have everything installed to start working right away. They can integrate with major GIT providers and with datasources.
π¨βπ» Why am I launching KeaML today?
So far, a close group of users have tested the product and provided feedback about it. Now, I'm launching the product to get my first customers, but mainly to get more feedback.
If you join the waitlist with the promo code PH6M, you'll get 6 months for free π₯.
I'm super excited about launching KeaML. I truly believe that's a tool that can help data science teams to work better. Can't wait to receive your questions and feedback! π
Hi @dinorah_margounato! Collaboration in data science teams has always been a struggle. Most of well known collaboration tools for software engineering doesn't work for data science. KeaML is trying to fix that!
Really love how the settings are immediately accessible, including Jupyter Lab, connections to well-known GIT
providers, and access to various data sources. Out of the park this one @gobaldia
KeaML seems to be a really useful tool, with a light and easy to use user inferface, from what I seen so far. Congratulations for the launch and I hope to be using it soon! π₯
Replies
KEAML
KEAML
KEAML
AirOps
KEAML
KEAML
KEAML
KEAML