Big-data analytics company Databricks Inc. is adding more automation, administration and security features to its Unified Data Analytics Platform today.

The updates are designed to make it safer for companies to leverage data for machine learning applications and other data-intensive tasks.

Databricks’ Unified Data Analytics Platform, based on the open-source Apache Spark big data framework, is used by enterprises to analyze data, build pipelines across siloed storage systems and prepare labeled datasets for model building. The idea is that organizations can then use the Unified Data Analytics Platform to train machine learning and other artificial intelligence models using their existing data.

Databricks said its software works well for these purposes, but the company is worried that leveraging data at such massive scale can be risky. The problem is that security systems are often fragmented, and corporate policies can be tricky to extend across an entire organization. That’s particularly true for companies using multiple cloud platforms to house their data, Databricks said.

To address these concerns, the Databricks platform now integrates what is essentially a fully managed security software-as-a-service offering, so it’s now possible to run Databricks clusters inside individual cloud accounts. In addition, Databricks now provides customer-owned data encryption keys and customized private networks to run each cluster, enabling customers to tailor the service for their own unique security needs.

On the administrative side, Databricks is adding more tools to support organizations that have thousands of users accessing their data. The new admin tools allow companies to audit and analyze all user activity within their accounts, as well as set policies for administering users, budget control, infrastructure management and more.

The last of today’s three big updates pertains to automation. It allows Databricks users to put analytics into production more quickly using a continuous integration and continuous delivery philosophy. To do that, Databricks’ platform is adding support for git repositories and application programming interfaces for tasks such as user management, workspace provisioning and cluster policies.

“Databricks is the only platform that has successfully achieved the massive scale and simplicity that enables enterprises to make data, business analytics and machine learning pervasive enterprise-wide,” said David Meyer, Databricks’ senior vice president of product management. “We’re committed to preserving this for our customers, regardless of if and how their cloud strategies evolve over time.”

Image: Databricks