DataRobot, a Boston-based automated machine learning company, made a slew of announcements this morning, expanding its platform to provide something new to both technical and nontechnical users. It also announced the acquisition of Zepl, which will provide DataRobot with an advanced development platform where data scientists can carry their own code. The purchase price was not shared by the two firms.
Nenshad Bardoliwalla, DataRobot’s SVP of Product, says his company aspires to be the market leader, and claims the path to that goal appeals to a wide range of consumer requirements, from those with little data science knowledge to those who can code machine learning in Python and R.
People appreciate automation, but they also want it to be adaptable. They don’t want just automation, but you can’t do anything about it if you don’t have it. They just want to be able to turn knobs and pull levers, according to Bardoliwalla.
Rather than creating a coding environment from scratch, it decided to purchase Zepl and integrate its coding notebook into the platform as part of a new tool called Composable ML. We are now offering a truly first-class environment for people who want to code, thanks to Composable ML and the Zepl acquisition, he said.
According to Crunchbase data, Zepl was created in 2016 and has raised $13 million to date. The company declined to disclose the number of employees or the purchase price, but the acquisition provides it with advanced capabilities, especially a notebook environment to attract advanced users to the platform. The Zepl features will be integrated into the platform, but the standalone product will remain available.
The Zepl acquisition, according to Bardoliwalla, is an extension of the automated side of the building, where these tools can operate in tandem with one another, with machines and humans collaborating to create the best versions. According to Bardoliwalla, this creates an organic mixture of the best of what a machine can produce using DataRobot AutoML and the best of what humans can do, with the aim of composing those together into something very interesting.
The company is also launching a no-code AI app builder that allows non-technical users to develop apps using drag-and-drop components from the data collection. It will also provide a feature to monitor the model’s accuracy over time. After a model has been in production for a while, the accuracy may begin to deteriorate as the data on which the model is based becomes outdated. This tool checks the consistency of the model data and alerts the team when it starts to go out of compliance.
Finally, the organization will unveil a model bias monitoring tool to aid in the detection of model bias that might inject racial, sexist, or other assumptions into the model. To prevent this, the company has developed a method that can detect when this occurs both during the model-building process and during development. It alerts the team to the possibility of prejudice and offers ideas for removing it from the model.
DataRobot was founded in 2012 and is based in Boston. According to PitchBook, it has earned over $750 million and is valued at over $2.8 billion.
