Skip to content

iko

iko enables teams and individuals to Get Data Products Right.

Work together with collaborative real-time Jupyter notebooks and see your teammates changes while they happen. Colleagues can either have read-only access, or edit access which makes pair programming on notebooks way easier.

Leverage GPUs and schedule long-running notebooks that survive closed tabs and network disruptions. You can close your laptop, go for coffee, and come back and watch the execution output as it's happening.

Load and manage data; train, validate, deploy, invoke, and monitor machine learning and deep learning models. Automatically track experiments to know which parameters produced which model on which data.

Enable non data-scientists to interact with notebooks without being overwhelmed by the code or the notebook interface. You can hide whichever cells you want. Simply publish AppBooks. Choose the parameters you want to allow people to tweak. This avoids exporting to PDF or HTML, which are not interactive. It also avoids building ad-hoc applications to showcase your work to clients or stakeholders.

Why

Real world machine learning projects almost always are the result of a group collaborating to solve a problem involving domain expertise. The current state of machine learning is in flux and involves many moving parts and a variety of tools that often become incompatible. Teams spend a lot of time dealing with low level details for data, development, deployment, and management, using ad-hoc ways to collaborate, and manage members joining and leaving the internal team or the client's team. That is for people lucky enough to have a team. The way projects get done is either by someone who can do it all, or when people can rely on others. Neither is affordable.

On the one hand, this time could be spent on solving the problem at hand. On the other hand, these problems have to be dealt with. Large organizations whose business relies heavily on machine learning may have specialized teams that can keep the organization at the state of the art when it comes to infrsastructure and tooling. The rest of the world does not have these teams, which means that the barrier to entry to leverage machine learning is high and requires considerable initial investment.

We're on a mission to solve these problems for ourselves, and for you. Read more about us.