Immuta Raises $20M Series B to Drive Operationalization of Data Science

We first encountered Immuta in August 2017, noting that the company was differentiated among the new breed of data catalog-centric data management providers by its focus on data science and advanced analytics. It was founded by former US intelligence agency contractors, so data security is also naturally a key focus. Immuta's initial positioning was on accelerating time to insight from data science projects through a virtual data layer that catalogs and provides access to existing data sources, thanks to the automated enforcement of privacy and policy controls.

While managing the input to data science projects largely remains its focus today, the company already has an eye on the next stage of its strategy, which is centered on the output of the data science efforts – specifically data profiling and machine-learning model monitoring and governance, the combination of which it refers to as 'algorithmic risk management.'

Immuta recently raised a $20m series B funding round to help finance the development of its algorithmic risk management functionality. The round was led by DFJ Growth, with participation from Dell Technologies Capital and Citi Ventures, as well as existing investors Drive Capital and Greycroft Partners – bringing the company's total funding to $29.5m.

Central to Immuta's longer-term strategy is its belief that managing data for building machine-learning models is very different from managing data for developing traditional applications. Specifically, it notes that data scientists require more ad hoc access to multiple subsets of data, which necessitates multiple data connections, while each system has unique metadata usage policies that need to be handled via custom logic. Immuta believes that as it sees greater adoption it will be able to leverage its understanding of metadata in relation to both positive and negative outcomes to provide recommendations and functionality to improve data profiling, model governance and model monitoring.

In addition to expanding its R&D and product teams, Immuta plans to use the funding infusion to grow its go-to-market team, establish inside sales and partnership teams, and expand into EMEA starting in Q3. The company currently has 32 employees (the majority of which are in engineering roles) spread across locations in Columbus, Ohio; College Park, Maryland; New York City; and Boston.

Immuta doesn't disclose its number of paying customers but has reported greater than 100% topline growth in 2017 and more than 500% growth to date in 2018 thanks to success in the federal government, financial services, life sciences/healthcare, and insurance verticals. As noted, that initial success has been based on offering a self-service platform for data scientists that enables them to use their preferred analytics and data science tools to securely access and collaborate on existing corporate data via an in-memory data virtualization plane that creates a catalog of all data in the data estate (including data storage, relational databases and the Hadoop data-processing framework, on-premises or in the cloud).

The company's platform also has functionality aimed at data stewards and data governance professionals, including the ability to audit data usage as well as dynamic policy enforcement. The platform's compliance capabilities were boosted in April with the launch of Immuta v2.1, which added the ability to generate automated governance reports based on the company's audit logs. The latest version also added Apache Spark support improvements, including native Apache SparkSQL policy enforcement.

 

Competition

There is no shortage of vendors providing data catalog-centric data management, including IBM and Informatica, as well as Alation, Waterline Data, Collibra, Cambridge Semantics, Unifi Software, Podium Data and Zaloni. However, all of these companies are primarily focused on data analysts and data preparation users, whereas Immuta is specifically focused on enabling data scientists and machine-learning workloads.

Hadoop vendors Cloudera, Hortonworks and MapR are also increasingly focused on data science workloads and catalog-based management of data in their respective platforms, but are more likely to be partners than direct rivals, especially since Immuta addresses the wider data estate, including data storage and relational databases. Given Immuta's data virtualization approach, potential customers might also be looking at the likes of Denodo, TIBCO and DataVirtuality, while its data masking and data governance capabilities might prompt comparisons with security players like BlueTalon and Dataguise.

With its plans to address model governance and model monitoring, we also think Immuta might encounter the new breed of data science management specialists such as ParallelM, Seldon, Metis Machine Datatron, DataKitchen and Hydrosphere.io. It could also vie with Domino Data Lab, SAS Institute, IBM, Talend and Hitachi Vantara.

That said, while all of these firms potentially compete with some aspects of Immuta's platform, the company is differentiated by its specific combination of capabilities.