FeatureByte, the developer of a characteristic engineering platform, has launched from stealth with a $5.7 million seed spherical. The Boston-based startup payments its SaaS resolution as particularly made for information scientists to simplify the creation, serving, managing, and monitoring of machine studying options.
A characteristic, also called a variable, is any measurable enter used for a predictive machine studying mannequin. As step one in ML mannequin improvement, characteristic engineering is the method of making use of area information (e.g., enterprise information, arithmetic, and statistics) to extract analytical representations from uncooked information. To be able to select probably the most helpful options for a selected mannequin, information scientists should choose, manipulate, and rework uncooked information right into a format that may be immediately consumed by ML fashions. There are many information preparation instruments made for the information analytics realm that may automate information preparation, however there’s a lack of automation instruments constructed particularly for AI mannequin workloads.
Given this information preparation limitation, characteristic engineering and administration is an intricate course of that may be gradual and costly. In line with, options are a number of the most extremely curated and refined information belongings attributable to how a lot time, effort, and talent is concerned with their creation. says that regardless of this significance, many organizations should not have an efficient characteristic administration system. Moreover, the corporate says that characteristic engineering is the weakest hyperlink in scaling AI as a result of it “requires the confluence of three distinctive abilities – area information, information science, and information engineering. Even in organizations with mature AI practices, these areas of experience reside in silos. And on the intersection of those silos lies a ton of friction.”
Fixing the issue of those disparate silos of experience is the purpose for FeatureByte co-founders Razi Raziuddin and Xavier Conort, each of whom are alumni. Raziuddin, FeatureByte’s CEO, scaled DataRobot from 10 to 850 staff and led its go-to-market technique. Conort, CPO at FeatureByte, was chief information scientist at DataRobot and constructed its R&D information science group. The startup’s $5.7 million seed spherical was led by Glasswing Ventures and Tola Capital, and the corporate plans to make use of the funds to scale its R&D and go-to-market operations.
“Our group has efficiently launched AI deployments for a whole bunch of organizations worldwide. Nonetheless, the one fixed problem enterprises face is characteristic engineering and administration. Xavier and I shaped FeatureByte to radically simplify the method for information scientists and software builders,” stated Raziuddin. “The market is extraordinarily fragmented, with siloed options addressing solely items of the puzzle. We’re creating an answer from first ideas to deal with full cycle that includes engineering and are excited to companion with Glasswing Ventures and Tola Capital to drive this imaginative and prescient and mission ahead.”
FeatureByte has plans for its cloud-based platform to have integration capabilities with Snowflake and Databricks. FeatureByte will likely be obtainable for early customers through an invite-only beta program. To be taught extra, go to.