Cuomo, a startup that provides an AI-powered platform to solve predictable problems in business, today announced that it has raised $18 million in a Series B round led by Sequoia, with participation from A Capital, SV Angel and several other angel investors. Co-founder and CEO Vanja Josifovski said the new funding will support Kumo’s recruitment efforts and R&D on the startup’s platform and services, which include data preparation, data analytics and model management.
Kumo’s platform works specifically with graph neural networks, a part of an AI system for processing data that can be represented as a series of graphs. Graphs in this context refer to mathematical constructs ends Connected by (also called nodes). Edges (or lines). Graphs can be used to model relationships and processes in social, IT, and biological systems. For example, the link structure of a web page can be represented by a graph where the vertices stand for web pages and the vertices represent links from one page to another.
Graph neural networks have powerful predictive capabilities. On Pinterest and LinkedIn, hundreds of millions of active users are used to recommend posts, people, and more. But as Josifowski points out, they’re computationally expensive to run — making them cost-prohibitive for most companies.
“Many enterprises trying to experiment with graph neural networks today are unable to scale beyond training data sets, which severely limits their ability to use these new algorithms that come with a single accelerator (the memory in a single GPU),” he said. TechCrunch in an email interview. “Through fundamental infrastructure and algorithmic advances, we have been able to scale data sets in the multi-terabyte range, enabling graph neural networks to be applied to clients with large and complex enterprise graphs, such as social networks and polygonal marketplaces.”
Using Kumo, customers can connect data sources to create a graph neural network that can then be queried with structured query language (SQL). Under the hood, the platform automatically trains the neural network system, evaluates it for accuracy, and prepares it for deployment into production.
Josifovsky says Cuomo can be used for applications such as new customer acquisition, customer loyalty and retention, personalization and next best practices, abuse detection and financial crime detection. Formerly the CTO of Pinterest and Airbnb, Josifovsky worked with Cuomo’s other co-founders, former Pinterest Chief Scientist Jure Leskovec and Hema Raghavan, to develop graph technology through research labs at Stanford and Dortmund University.
“Companies spend millions of dollars storing terabytes of data, but can only effectively use a fraction of it to generate the predictions they need to drive forward-looking business decisions. This is due to gaps in core data science capabilities, as well as the significant time and effort required to successfully bring predictions to production,” Josifovsky said. . We enable companies to move predictive analytics from minimal resources to an easy-to-write SQL query, enabling predictive analytics to become virtually ubiquitous – to fit a wide range of use cases. In a very short period of time in the organization.”
Cuomo remains in the pilot stage, but Josifovsky said he has “more than a dozen” early adopters in the organization. To date, the startup has raised $37 million in capital.
