Union.ai, provider of the open source workflow orchestration platform Flyte, announced they have closed $10 million in seed funding led by NEA with additional participation from a select group of angels to accelerate the growth of Flyte and provide even greater assistance to its user community. The investment will be used to further grow the Flyte open source ecosystem while building out a commercially available Union Cloud platform.
Machine learning (ML) has the potential for explosive impact across all industries. Existing tooling has not caught up to modern ML requirements, including reliability, reproducibility, enforcement of quality and efficiency. This undermines the efficacy of the models and reduces potential ROI.
“The requirements for ML/Data are very different from delivering services to production,” said Ketan Umare, CEO at Union.ai. “ML/Data workflows are focused on experimentation and are stateful. Meanwhile, dealing with infrastructure procurement is one of the last things a data scientist wants to do. With Flyte, we are solving common issues for running ML workflows, including separating the concerns of infrastructure management from the end user, providing guaranteed reproducibility (a common requirement for deploying complex ML and data applications) and allowing organization-wide re-use of existing applications. As a software engineer, I do not remember the last time I wrote a sorting algorithm, but in the world of ML and Data, redoing the same thing over and over again is the norm. We want to change that.”
With Union.ai, engineers and data scientists can abstract away the complexity behind workflow automation and focus solely on developing models to solve their core business problems. Union.ai offers new services on top of Flyte so that users can integrate with state of the art machine learning tools, develop reproducible and shareable models, and reap the benefits of a thriving open source, serverless platform.
“When we started thinking about how we could further help the community, we realized that offering a hosted solution where users could get up and running in minutes would help improve their velocity,” added Ketan. “Many teams expressed they were reluctant to embark on hosting Flyte on their own. We want to reassure them that we’re like ‘Flyte’ attendants, here to help them every step of the way through a cloud offering that allows them to start building production-grade ML pipelines in minutes.”