Activeloop, a Y Combinator company, has raised $5M to help data science teams create, store, and collaborate on datasets for deep learning applications.

Activeloop, a Y Combinator company, has raised $5M to help data science teams create, store, and collaborate on datasets for deep learning applications. The seed round is led by 468 Capital and CM Ventures with participation from Tribe Capital, Shasta Ventures, and angels like Tobias Knaup (CEO of D2IQ) and David Aronchick (co-founder of Kubeflow).

Activeloop’s list of investors also include the creator of Gmail, Paul Buchheit, SmartGate, and AiSprouts, among others.

Currently, around 31% of ML projects fail due to lack of access to production-ready data. Activeloop aims to solve this problem by delivering a unified data framework to enterprises that allows data science teams to create, store and collaborate on production-ready AI native datasets.

Marketing Technology News: Enrichment TEAM Engages in a Strategic Collaboration with HokuApps to Accelerate Digital…

“Activeloop provides a unified data stack that should be an integral part of any modern machine learning infrastructure”

— Tobias Knaup, CEO at D2iQ.

Its open-source package, Hub, is already being used by researchers from Google, The University of Oxford, Red Cross, and innovative computer vision startups and enterprises. Hub has also trended #1 in Python worldwide among 3.7 million repositories on GitHub.

Recently, Andrew Ng started the Data-Centric AI movement, emphasizing that data is food for AI. He encouraged AI practitioners to bring a more systematic approach to working with data.

Marketing Technology News: MarTech Interview with Ron Jaworski, CEO at Trinity Audio

“We are glad to witness the rise of the data-centric AI movement. Activeloop has been empowering machine learning engineers to build a solid data foundation and work efficiently with data at scale since day one,’’ says Davit Buniatyan, CEO of Activeloop. ‘“That is why we are building a standardized database for AI to help enterprise teams double the speed of ML development and cut their infrastructure costs by at least 30%.’’

“As machine learning datasets are exploding in size and complexity, it is quickly becoming untenable to rely on homegrown data management tooling,” says Tobias Knaup, CEO at D2iQ. “Activeloop provides a unified data stack that should be an integral part of any modern machine learning infrastructure.”

The next decade of development in machine learning will focus on data rather than models. Activeloop aims to be the canonical tool for this data-centric future.