ikuzo

Inspiration

Too many DS enthusiasts without SWE skills who don't know how to serve their ML model.

What it does

Allow DS enthusiasts to serve ML models online from the notebooks using just 3 lines of code.

How we built it

• Using Ruby on Rails to create a web backend and frontend that spins up new APIs, and deployed on Heroku
• Created a Python library usable in Jupyter notebooks for Data Scientists.
• Used pubnub.com for messaging between the notebook and our product.

Challenges we ran into

• It doesn’t work if the scientist is hosting the notebook on his local computer and closes it, but it’s not a problem if hosted on the Cloud
• To understand the data scientist experience and making the usage of our library to be as easy as possible

Accomplishments that we're proud of

• We built a functional end to end Proof of Concept for models running on text input and output

What we learned

this only the beginning.

What's next for ikuzo

• Finish up the UI
• Get the first 100 users
• Save people tonnes of time