1 of 10

Turning AI Deployment Friction Into Product Confidence

ENP74 – Human Factors Capstone

Professor: Linda Borghesani

Team: Josiah deGrasse, Christie Hao, Joey Marmo, Marlon Ward, Nancy Yang

Sponsors: Beau Morley, Yahav Manor�

with Red Hat OpenShift AI (RHOAI)

​

May 7, 2026

2 of 10

Our Team

2

PM + UX Writer

Christie Hao

Josiah deGrasse

Joey Marmo

UX Researcher

Nancy Yang

Marlon Ward

UX Researcher

UX Designer

Client Liaison +

UX Designer

3 of 10

Red Hat helps companies run software at enterprise scale

  • Red Hat builds open-source technology for cloud and enterprise software
  • Red Hat OpenShift AI (RHOAI) helps teams build, test, and deploy AI models into real products
    • Build and test AI models
    • Scale AI work into production
  • RHOAI User persona: AI engineers who deploy AI models on RHOAI daily

3

Red Hat team wanted to make the deployment process for AI engineers feels “one-click” for speed and ease.

4 of 10

4

5 of 10

User interviews indicated frustration around system guidance during deployment

5

Quotes

​

"The deployment failed, I had to switch between consoles just to understand why. You're reading logs in one place, and checking configurations in another."

-User Interview

​

“I don't know what's on this cluster, I don't know if this model is suitable for my hardware. That's when I go ask Claude/Gemini."

-User Interview

​

"Would like the option to still be able to customize further than just the small, medium, large options."

-User testing

8 user interviews

Across RHOAI expertise:�5 without RHOAI experience, 3 Red Hatters

3 pain points

No feedback during failed deployments

Technical terms are inadequately explained

Some preferred earlier versions for certain settings

6 of 10

Concept Testing

ROUND 1 OF 2

WHO WE TESTED

4 Red Hat users — Platform Engineers and Researchers. In-person at Red Hat’s Seaport, Boston office.

HOW

  • Interactive Figma prototype
  • Think-aloud protocol
  • Likert scale ratings (1–7)

NEEDED MORE WORK

  • Actionable fixes, not just “likely causes”
  • Logs for all deployments, not just failures
  • Hardware: resource limits on summary page

Troubleshooting sidebar

“This is already a million times better from a UI standpoint.” — Platform Engineer

Log consolidation

“I love this, I don’t have to go between the console and the OpenShift UI.”

— Platforms Engineer

Status display

“Status for the model is the best improvement made.”

— Platform Engineer

Contextual tooltips

“There is value to the link to documentation.”

— Platform Engineer

7 of 10

User Feedback

ROUND 2 OF 2

WHO WE TESTED

5 Red Hat users — UX Designers, Product Managers, and RHOAI leadership.

Online in a Zoom meeting.

​

GOAL

Validate the refined prototype.

​

KEY CHANGES MADE

  • Added logs for all deployment states
  • Actionable troubleshooting
  • Sticky tooltip popovers with doc links
  • Preset + full customization for hardware

“It makes me feel good because it supports the same things I found in my own research over the last two years.”

RHOAI Leadership

Eliminated tab switching “Fixed biggest issues.”

Actionable troubleshooting “I like to know the actions I can take from here.”

Reduced learning curve “I find this very valuable to link to documentation.”

Streamlined hardware config “I like the option to customize further while I have the presets.”

8 of 10

Prototype Demo

v2.0

8

9 of 10

Outcomes

9

    • Validated that RHOAI users need more guidance during deployment

Validate

    • Identified concrete improvement areas:
      • Deployment failure troubleshooting support
      • In-product explanations

Identify

    • Enhanced deployment confidence for AI engineers

Enhance

10 of 10

Next steps

10

Conduct usability testing with participants representing a range of Red Hat OpenShift AI expertise levels

Update Red Hat documentation for relevance and specificity

Coordinate with Red Hat product leadership on implementation planning