.png)
Deepchecks was founded by a group of geeks, that lived and breathed machine learning before it was a thing. We were lucky to have the chance to lead top tier machine learning research groups, and work on dozens of different machine learning tasks, involving many different types of datasets and business constraints.
Through our own experience, we learned something very interesting about machine learning systems. Even though our organizations were spending millions of dollars building ML models, we realized that we don’t have a proper way to check if something is wrong with them. And unlike most classic software systems, we learned these systems can fail silently, and that certain issues can go undetected within a ML system for years.
As we began to deploy more and more models into production, we were astonished to discover that there was nothing out there that could help us with this huge problem: Neither tools, nor literature, nor good advice. We would have loved to buy an out of the box solution, that would continuously test our ML models and let us know about the issues that arise. We knew that our organizations were a bit ahead of the curve, but we realized that every company in the world will probably have ML systems in production within a number of years.
So we set out to build the company that we sought and didn’t find: Deepchecks.
Website: www.deepchecks.com