HPE Ezmeral (Unified Analytics)

Project Overview

HPE Ezmeral Unified Analytics is an enterprise data platform that unifies data management, AI, and analytics in a single solution. The platform provides transparent secure access for AI workloads, analytics, and data lakes while enabling consistent data access across diverse environments.

Often as a designer or developer, we work on existing products and codebases and it can be rare to get to start from scratch. As UX Engineering Manager, I joined HPE Data right as Ezmeral was getting started. Initial designs and some prototypes were in early stages, and I had the opportunity to complete the team by hiring the remaining designers and engineers.

Together, we spent about a few months building out a Storybook and infrastructure for the app. We had the strong foundation of the HPE Design System, but built out complex components, tables, patterns and templates suitable for our design approach. Meanwhile, my designers formalized design assets and tested the initial application features.

I led cross-functional teams through the complete product development cycle, and over the year we successfully built and shipped Ezmeral to customers across 11 countries.

Technical Tid-Bits

• React-based frontend with Storybook component library
• Kubernetes-native containerized architecture
• Support for multiple data formats (HDFS, S3, NFS, POSIX)
• Integration with Apache Spark, Presto, and other analytics engines
• Enterprise authentication and authorization systems

Challenges

• Balancing comprehensive user testing with aggressive timeline
• Building a scalable UI framework for complex data workflows including a native SQL editor
• Ensuring consistent user experience across diverse data sources and integrations with varying setup requirements (lots of magic happened behind the scenes to keep the UI experience consistent)
• Meeting enterprise security and compliance requirements
• Coordinating international design and development handoffs

Key Learnings & Impact

We delivered a robust product, but our heavy emphasis on in-app workflow testing came at a cost. By focusing almost exclusively on perfecting the application experience, we delayed our market entry and overlooked a critical first impression: installation.

Early deployments required extensive on-site engineering for custom configurations and workload tuning. It was time-consuming, resource-heavy, and frustrating for customers. This led us to pivot in our second release, replacing full custom multiple day installs with a simple onboarding wizard offering 5–6 pre-configured profiles that could be up and running in a few hours

The shift transformed setup from a complex, fully customized process into a streamlined "appliance-like" experience. It was fast to deploy, easy to manage, and still flexible enough for most customer needs. It was a pivotal lesson in designing for the entire customer journey, not just the app itself.

Read more about the install project here.