All case studies

Craft Your Own AI Dashboard

Designing a holistic, consumer-grade experience for no/low-code AI/ML app creators — not just expert data scientists.

My Roles

UI Design

UX Design

Design Discovery

Design System

Enhance your preventive upkeep program with predictive maintenance

Challenges

A Lot of Great But Unconnected Experiences

Uptake was instantaneously great at showing value to its customers who wanted to monitor their assets and make predictions as to their health. It got started by servicing individual industrial verticals (wind, rail, fleet, etc.). The issues was that there were several very unconncted experiences within the suite. Additionally, it was realized that there was an appetite not only for us to provide curated dashboards and templates, but for users to customize and even build their own monitoring apps.

Objectives

Get Organized and Make an App That Builds Apps

Create something consumer-facing that was compelling, approachable, and attractive to a tech-savvy individual who wanted to build their own app at any level of expertise. Customization was crucial for analysts who needed to focus on the right problems.

Improve the network of services Uptake provides into a more user-friendly experience.
Create a low/no-code app builder that can render live machine data.
Surface the value of Uptake's specialty: predictive analytics.

My Role

Combine The Past and Propose The Future Visually

I was a senior visual designer on the platform team, tasked with a blend of consumer initiatives — onboarding, broader app architecture, and implementation of the App Studio. I was deeply embedded with a strong engineering team where we were encouraged to think big and rapidly stand up a new POC version of the Uptake experience.

Application Users

A P P B U I L D E R S

Senior Data Scientist · Regional Solar Operations

Dana has 8+ years in predictive analytics and lives inside Uptake's Data Science Studio all day — building, tuning, and deploying models across a large fleet of industrial assets. She's the go-to expert on her team for model health and knows the platform inside and out. Dana wants to monitor model performance and drift across dozens of deployments while retaining the ability to deep-dive into raw sensor signals and anomaly patterns at any time. She needs to configure custom alerts and thresholds for specific asset classes and share findings with ops teams without losing any data fidelity in the process.

Operations Manager · Mid-size Wind Company

Marcus runs a team of 12 responsible for keeping 200+ trucks on the road. He's tech-savvy but not a data scientist — comfortable with software tools and has a clear, practical sense of what his team needs to see to make fast maintenance decisions. Marcus wants to build a custom monitoring app without needing to pull in engineering resources every time. He needs a clear view of predicted failures before they cause downtime, visibility into which assets are at risk this week versus next month, and a simplified view he can share with his maintenance crew.

Outcomes

It All Happened So Fast

The pace was some of the fastest I've ever experienced. We were all focused on the goal ahead and making measured improvements over chasing perfection — trying to make hard things feel easy. It was extremely rewarding.

A C H I E V E M E N T S

Rapid App Studio launch: We shipped the App Studio in sprints, not quarters or years
Immediate user feedback: The App Studio was adopted right away, generating a steady stream of feedback back to the team

UI contributions to the design system: Drawing on my background in the design system, I was able to build new components and contribute them back for refinement
Real-time dashboards and predictions: Time-series charts, threshold bands, status badges, histograms, and sparklines — all customizable by users

H U R D L E S

Less time for research: We operated on a lot of assumptions to get the Studio off the ground
Building an app builder: The meta experience of creating an app that creates apps was genuinely complex

Real-world testing reveals the truth. Applying the system to real scenarios exposed gaps and helped embed industry expertise into UX and data science.
Designing for diverse roles is challenging. We built flexible experiences that support users ranging from executives to field technicians.
Accessible design takes iteration. Balancing clarity, usability, and accessibility requires careful design.
Make measured steps. It is very easy to become overwhelmed when you need to design everything.

AI wasn't always trusted: We had to provide ways for expert users to validate AI insights due to a general lack of trust in the outputs
Mixing UI paradigms: Some early experiments were visually jarring — light and dark themes were used interchangeably without clear purpose

User Journey

Discovery

A senior interaction designer and I were assigned to design the App Studio experience and explore how we could visually improve onboarding. We jumped straight into wireframing key milestones that the team knew users were asking for. Everything needed to be strung together into a clearer, more cohesive experience. We rapidly iterated toward what we felt was the simplest, most user-centric flow through the App Studio suite — building a hub to anchor all the other experiences.

Put an ugly handdrawn pic here

Improving Onboarding Experiences

A New Login Screen

Some of the proposed visual directions. We also A/B tested to see whether users preferred more resources and information upfront.

The winning design was clean and simple. We found that our expert users weren't exploring different verticals of the platform until they were already logged in and actively creating.

Creating a Simpler, More Friendly Hub

I was asked to create an "ideal" state for the app suite.

The Application Studio

These are screens I designed to guide users into the App Studio.

As users started building more apps, feedback revealed that they liked thinking in pages — so we created a list view for monitoring each page within their new app.

Page Creation Modals

Version 1: Proto App Studio

Creating a builder canvas one step at a time — a single-widget POC. We were primarily testing the best approach to rendering individual widgets in real time: How hard was it to provide data channels, connect them, and render?

Version 2: Thinking in Slots & Config Panels

Once we could render individual widgets, it was time to think in slots, measure performance across multiple rendered components, and build out a configuration panel on the right.

From there, we added the rendered chrome of an individual app page (header, subheader, background, etc.). This was optional, but many users appreciated seeing the full page visually.

We then added the ability to customize app styles, make all components draggable, and preview the result.

Expert App Builder

Evolving to a Robust Canvas Experience

As the App Studio matured, a new need emerged. Data scientists wanted more freedom to customize components and a more open way to lay out their workflows and data science pipeline journey maps. This led us to design a more fluid canvas experience.