Contrapuntal AI | R-Shief

Sole UX/UI Designer · Wireframes, prototype and design system · Figma

Case Study

|

Product Design

|

Launched

|

Case Study | Product Design | Launched |

Overview

Contrapuntal is a human-centered AI platform for public research, built by the Creative Critical AI Undercommons at UC Santa Barbara alongside R-Shief, a 15-year social media archive. Where most AI tools hide their sources, Contrapuntal keeps every answer tied to its citations, so researchers can see where knowledge comes from.

I joined as the only designer on a cross-disciplinary team of faculty, graduate researchers and engineers, and owned the experience end to end: style guide, information architecture, every screen and the component library the developers built from.

Outcome

Launched publicly in September 2025, featured by UCSB's Division of Humanities and Fine Arts, and in grant conversations with Netflix and other companies to fund what comes next.

The challenge

The brief: transform R-Shief from a static informational site into a living, feed-driven platform, using X and Hugging Face as reference points, with Contrapuntal's AI built into the experience. The original scope was a style guide and seven pages. The real challenge was making dense research data feel as approachable as a social feed, for scholars and the public alike, without losing the depth researchers need.

Designing from user stories

The team wrote scenarios for why a researcher would choose Contrapuntal over ChatGPT or Semantic Scholar. Two patterns kept surfacing, and nearly every scenario described thinking in webs of related ideas. Those insights became the backbone of the product.

Text-first researcher

Starts from a source and traces its connections

Led to dataset pages with labeled features and a path into related texts.

Question-first researcher

Starts from a theme and explores outward

Led to a prompt-first entry point and an interactive network graph that makes relationships clickable.

What I Designed

01

A prompt-first front door

A single input invites people to start with a question, not a menu. I designed the full flow from request to loading to response, plus a persistent assistant users can open from any page so the AI is always one click away.

02

A dataset library built for discovery

Researchers filter by source, format, size and modality, open detail pages, and create their own datasets through a guided flow. My dataset dashboard design was the screen UCSB chose to feature in its launch coverage.

03

Making connections visible

The citations dashboard pairs an interactive network graph with citation cards and linked datasets, turning the team's "webs of ideas" into something people can see and click through.

04

A social layer and a system to scale it

A Data Stories feed, follow suggestions, trending datasets and profiles turned a static site into a community. Every screen ships in light and dark mode, built on a component library I designed on a Bootstrap foundation so it mapped directly to the engineers' code.

Weekly check-ins with the project lead, milestones in Notion and day-to-day collaboration in Slack. After handoff, I stayed on as the design point of contact while developers built page by page from my Figma prototype.

Process

May 2025
Kickoff

Presented style guide and mockup direction at the team retreat.

May to June
Design sprint

Landing pages, app flow, core pages, then final designs.

July to August
Build

Design support for a 5-person dev team.

September 2025

Public launch

Live at contrapuntal.net

November 2025

Recognition

Featured by UCSB, with grant talks underway

Results

In the press

UCSB's Division of Humanities and Fine Arts covered the launch, crediting me on the platform's Developers and AI Ethics team and featuring my dataset dashboard design. After launch, the project lead was interviewed by Netflix and other companies for grants to continue the work.

7 to 30+

Scope grew from seven briefed pages to 37 screens and flows

2 themes

Light and dark mode designed for every screen

Shipped

Built from my Figma and launched publicly in September 2025

What I took away

Designing for a team of scholars and engineers taught me to translate between very different priorities and turn them into one clear experience. Matching my components to the developers' framework from day one meant fewer handoff questions and a build that stayed true to the design. It's the same approach I bring to ecommerce: own the experience, and make it easy to ship.