Kaskada, machine learning startup, closes Series A funding

Startup Kaskada is developing software to help integrate the roles of data scientist and data engineer in Machine Learning workloads being deployed more widely in organizations. (Kaskada)

Machine Learning (ML) startup Kaskada announced Tuesday it has completed a Series A funding round for $8 million from Voyager Capital and others, bringing its total funding to $9.8 million.

The Seattle-based company, founded by Google Cloud and AWS engineers, is working on a software platform for feature engineering to help unify the ML process across enterprises and other organizations.

The company claims it can help organizations make better predictions and speed innovation by integrating data science and data engineering workflows. Kaskada makes software for feature engineering and feature serving, which includes an interface for data scientists and an infrastructure for computing, storing and serving features in production.

Features are defined as independent variables used in ML algorithms. A feature is a kind of input that is an individual measurable characteristic of a phenomenon used in pattern recognition and classification workloads.

“Most people focus on models and training models in ML and forget feature engineering, which is key to getting success,” said Kaskada CEO Davor Bonaci, in an interview. He previously worked at Google Cloud on Cloud Data Flow.

Kaskada software is “trying to redefine for both the data scientist and data engineer a system that sits in the middle that they both work with,” Bonaci said. “The system is a kind of Application Programming Interface.”

Each party uses the system instead of the data scientist developing data sets that are tested and then used by the data engineer. “Think of the data scientist and the data engineer working like the riders and drivers in Uber,” he explained.

In today’s organizations, the roles of data engineer and data scientist “work inside their notebooks, and this is a really long process…The data scientist takes several weeks to turn the work over to the data engineer and we are trying to get that process down to the push of a button.”

Kaskada’s software will run on standard hardware in a data center. Bonaci claimed Kaskada’s product can do the work provided traditionally by 10 or more engineers.

The technology can be used in a variety of recommendation and ranking systems for dynamic pricing models, fraud detection, Internet of Things end user devices and any product where a company wants to personalize experiences. “What is really changing with big data is that companies are delivering personalized experiences that are more tailored to your interests and what you want,” he said. “That’s the next stage in every industry.”

Bonaci said the Kaskada software is “quite novel” although other companies are attacking the same problems. Kaskada’s system operates in the same vein as Uber’s ML platform called Michelangelo, first introduced in 2017, “but we have gone several steps further,” Bonaci added.

Kaskada plans to use the Series A funding to expand its software engineering team, now less than 20 in size. The first product will be delivered sometime in the first half of this year. In addition to Voyager, the Series A funding comes from NextGen Venture Partners, Founders’ Co-op and Walnut Street Capital Fund.

James Newell, managing director at Voyager, said Kaskada’s software addresses a big market need. “Nearly every company today is pouring significant resources into their data science efforts, and very few are seeing results that meet their expectations,” he said in a statement.

RELATED: Fastest growing tech job roles: engineer/scientist in data or ML