Client: IBM Hybrid Cloud, Analytics Platform, Private Cloud
My time working as a UX researcher and designer on the Analytics team has been focused on defining user needs within the feild data science at a corporate level, ideating on the problems data scientists face working with secure data on-premises, and designing low and high fidelity wireframes and user flows for feature updates. Our team works primarily with our development and offering management teams to deliver the updated experience for Data Science Experience Local on Hortonworks Data Platform.
DSX won a Red Dot this year for Communication Design in the Interface & User Experience Design category, as our team works on the next generation of the product, designed for businesses that work on a private cloud infrastructure.
UX Design, User Research
Polly Adams, Rich Kam, Renee Mascinaras, Jason Azares, Zoe Padgett,
Carrie Curtin, Nina Dang, Elika Berlin
Leadership: Caroline Law, Jessica Gore
On a weekly basis, I create research artifacts, storyboards and wireframes for new features and flows, as well as screens and click-through prototypes of DSX Local to share with our development team and research participants.
Any data scientist will tell you that analyzing data to better understand complex subjects is not an easy task. There are a myriad of tools that may or may not work together, it’s difficult to get access to data, and setting up the various back-end systems is a real chore. On top of this fight against technology, they tend to feel alone in battle. In the exponentially evolving inudstry, they want to work with other data scientists, to learn from each other, share algorithms & approaches to analyzing data, and to publish the results of their efforts. They want to work with peers, too- data engineers who can help them prepare data, business analysts who will adapt their results into new approaches.
The design for IBM Data Science Experience (DSX) arose directly from these insights into data scientist needs. This understanding led to the primary user experience innovation: the convergence of a collaboration site with popular open-source tools to explore the data. In one place, users can start by learning from the projects of other data scientists as well as fork the available work into their own explorations. They can publish their work if it’s of public benefit, or share with peers if it’s confidential. DSX is free to the public, but companies can purchase additional capability as needed.
This is where my team sat, designing new features and integrations on Data Science Experience Local. My team crafted solutions for companies operating on their own private cloud infrastructure to give them the same experience with more capabilities, like easily accessing enterprise data sources quickly without needing to move data outside their firewall.
I was trained at IBM Design to be a User Researcher for my role on my team in San Francisco, though I eventually transitioned to a product design role. While getting aquainted with the team, I mainly conducted qualitative research through interviews, contextual inquiry, and spent time working on synthesis from our teams findings and client workshop artifacts.
After our partnership with Hortonworks was established, we sent over a formal introduction to the personas we had worked on to build a common understanding and empathy for their day-to-day tasks. I helped redesign many research assets and create a high level walk through of the information we had gathered cumatively over the last year of working on DSX.
As deadlines approached and our development team released new updates, my role as a researcher on my team shifted to building prototypes with InVision and remotely testing these with participants. My deliverables from this time took the shape of presentations of findings to share with my team and actionable lists of preferences from studies focused on interviews and heuristic evaluations of the current builds from development.
Through out the time I spent as a researcher on our team, I was able to conduct and lead a variety of research initiatives from both a qualitative and quantitative standpoint. Majority of my time was spent turning whiteboard exercises into diagrams for our personal use, as well as for presentations and deliverables to development and research participants.
This particular example below (click to enlarge) examines the current experience and pain points ("As Is") of an enterprise company procuring new software through the IBM Analytics platform, and the idealized future vision ("To-Be") of what it could be like with our solution to unify the Anyltics Platform.
The definitive persona work I had a hand in directly impacted design decisions to enhance the experience of DSX Local and drove my personal understanding of the product as I shifted my focus to design work. I asked to take on more product design work to gain software knowledge and get my hands dirty in UX design and systems architecture again. I am still deeply passionate about the need for user research within the product design cycle. I am also endlessly grateful for the opportunity to learn more about research methods from some of the most intelligent and organized people I have ever met, and excited to take some of the lessons I've learned into the rest of my life and career.
My pivot to UX design wasn't easy. I had spent months flirting with the idea, and knew my way around Adobe products, Keynote, and InVision... but I was still pretty new to Sketch, the tool our team uses the most for designing prototypes. I spent the first few months working on that learning curve, while getting more practiced in the side of UX design I felt familiar with: wireframing and ideating on how to improve the experience.
Right away, I started spending my days evaluating existing user flows and how they might need to change with added capabilities and features. Borrowing some task analysis tools I had familiarized myself with as a researcher, we were able to take a step back and really define our user's problems and a ideate on elegant solutions working within our framework, rather than overstuffing DSX Local with every new features our clients were asking for.
A lot of the work that our team has done on DSX Local is still not publically available, but below are a few examples from the feature updates that made it into our last December release. I'd be happy to share more examples of my design work with you upon request.
While DSX Local has very similar functionality to DSX in the public cloud, there are a few key differences in functionality:
1. The Community content in DSX local will not update continually as does content in the public cloud
2. DSX Internal Connections replace the Connectors to external data sources available in the public cloud.
3. Users are focused on deployment as well as the exploration that is promoted in DSX Desktop on the public cloud.
While taking these differences into consideration, my squad focused on the updates we would be making to enhance the experience of DSX Local within projects. The goals I met (below) within my team were to simplify project creation, asset navigation, and the process of deploying and tracking model accuracy from a project.
Putting the "Create" drop down in the action bar allows users to upload and set up projects, notebooks, and import assets with other file types during any stage in the process, and connect them to a specific project for future use.
The main goal of this feature redesign was to enable users to create a new project from anywhere in DSX, adding the feature to the actions bar. My role in this was largely focused on the design of the project detail page, defining the meta data that's important for users to see at a first glace and which actions should be available at this level.
The overview page also includes hot links to important aspects of each project, like the working assets, collaborators, and connections to data sources. Users can easily access the detail page for each of these features and sort through bookmarks to find what they're elooking for more quickly.
Users working within projects in DSX may need to utilize hundreds of different assets to build and deploy functioning models. For larger coporations with large, secure data sets and information stored on their private cloud, navigation can get a bit tricky. Our redesign focused on enabling users to sort through their pool of assets based on their existing expectations and preferences for finding assets on their local computers.
DSX Local introduces the functionality to toggle to a file directory structure while exploring assets in a project, offering an alternative and familiar visual format for file searching. The concept allows users to toggle back and forth between views while "keeping their spot" in terms of larger file location.
Data scientists and other users of DSX on desktop or the public cloud are more focused on experimentation and exploration, publicly sharing and discussing their work with collaborators. DSX Local users work with resources on-prem, and their work is not as accessible to others. Once a model (algorithm to deetermine insight based on data) has been tested and created, users must deploy it for function within their organization. DSX Local enables high level users (a cheif data scientist or administrator) to focus on model management and health at a glance.
Once a model has been deployed, it's "health" refers to the accuracy it retains in use. A model management dashboard where users can oversee general health, take down or deploy existing models, as well as publish models and assets to the internal community page for their org.
More to come after the GA of the model management dashboard in DSX Local.
I'm incredibly thankful for my time at IBM. Between my 3-month training in Austin and deployment to my team in San Francisco, I've had the chance to get to know and work with some tremendously talented people, learn to embrace ambiguity and complexity in the thick of real world product releases, and dip my toe into the tech industry. I've even gotten to see our CEO a few times, eloquently explaining industry-first initiatives in quantum computing and machine learning with the elegant simplicity she might use to explain her signature headband. More than anything, IBM has given me the confidence to believe I can play a role in collaborative problem solving, ship real solutions, and (above all) think.