AI - Powered Research & Data Visualization Interfaces
Three AI-powered interfaces exploring what institutional archives become when reorganized around relationships instead of records. Developed in dialogue with the Harvard Art Museums, MoMA, MOMA PS1, and the Institute of Black Imagination.
Designed and built three AI-powered research and data visualization interfaces, using Claude-based data pipelines to transform cultural datasets into relational maps, concept taxonomies, and interpretive layers; owned concept, UI/UX, information architecture, prompt design, data structuring, and Cursor-assisted development, with two prototypes deployed.
What Else Was Happening?
An AI-powered research and museum discovery interface for the Harvard Art Museums collection.
The project begins with one object and opens the collection outward into a field of relationships. Using Harvard Art Museums API data and Claude-powered relationship mapping, each object becomes the center of a visual constellation — showing what else was being made, used, traded, or imagined around it.
Built independently: Concept, research, data structure, AI pipeline, interface design, frontend development. Dataset: 300+ Harvard Art Museums objects
Tools: Harvard Art Museums API, Claude API, Claude, React + Vite, Cursor, JSON pipeline
Echo exhibitions
An AI-powered data visualization interface that maps 25 years of MoMA and MoMA PS1 exhibition history through recurring curatorial themes.
Exhibition histories are published as chronology — one show at a time, no view of the thinking underneath. Echo Exhibitions tags twenty-five years of programming with a taxonomy of 19 categories and 134 curatorial concepts, then makes the archive analyzable: browse any concept, chart how often it appears year by year, and overlay themes on a shared timeline to compare how the two institutions think differently about contemporary art. Scraped, extracted, and normalized with a Node.js pipeline and the Claude API.
Built independently: Research, scraping pipeline, taxonomy design, Claude concept extraction, data normalization, UI/UX
Dataset: 1,125 exhibitions, 2000–present, MoMA + MoMA PS1
Tools: Node.js, TypeScript, Cheerio, Axios, Claude API, React + Vite, Cursor
IBI: Memory, Position, Future
An archive interface that reorganizes 104 Institute of Black Imagination podcast conversations into three analytical layers of intellectual movement.
A podcast archive is usually a feed — 104 conversations, browsable only one at a time. This interface reorganizes the full archive through three analytical lenses: what each guest inherits (Memory), what they push against (Position), and what they reach toward (Future), with every episode linked to others sharing a specific thread. A chronological feed becomes a map of collective intellectual movement.
Built Independently: Research, prompt design, episode analysis, relationship mapping, interface design, frontend
Dataset: 104 podcast episodes
Tools: IBI RSS feed, Whisper, Claude API, Playwright, React + Vite, Cursor