visualization fragments
Collected, classified, and made explorable in the open Vis-Sieve demonstration.
Researcher, designer, educator, and engineer
Visual systems for complex questions.
I build interactive media systems that help people examine large, messy, or contested bodies of evidence.
My work connects communication research with visual design, software engineering, and critical inquiry.
A few concrete outcomes
What the work had to handle.
browser payload
A 98.6% reduction using DuckDB and Parquet, while keeping exploration client-side.
dimensions and features
A theory-guided persuasion framework whose individual measurements remain inspectable.
About
I move between research questions, visual design, and production systems.
I am an emerging-media researcher, designer, educator, and engineer based in Huntsville, Alabama. I earned an M.P.S. from NYU's Interactive Telecommunications Program and recently led development for NYU's Colonial Networks while contributing to interdisciplinary research in visual analytics and interpretable AI.
I currently work in engineering operations for high-volume automated manufacturing and continue building research software across visualization, digital humanities, and language technologies.
My work spans interactive media, data visualization, research software, and human-centered AI, with a focus on how computational systems shape interpretation and communication.
01 / Selected work
Research questions, made explorable.
Projects that connect rigorous data work with clear, responsive interfaces.
35,000 fragments2D + 3D exploration
Vis-Sieve
An end-to-end multimodal pipeline and interface for classifying and exploring large collections of scholarly figures.
- Linked 2D and 3D views built with D3, Observable, Three.js, and WebGL
- An end-to-end workflow from figure extraction and multimodal classification to visual exploration
- Automated labels evaluated against human judgments and published as first-authored research
Illustration above. Open the demo to see the working interface.
OBJECT1775Paris
OWNERRecord 08contested
PLACESaint-Dominguelinked source
Provenance Stories
Nine interactive art-history narratives connecting artworks, ownership, place, colonial wealth, and contested evidence through linked text, timelines, maps, and networks.
My role: Lead developer and research technologist, from data model and ETL to interface design and production deployment.
Illustration above. Open the stories to see the live narrative interface.
Radar overview15 dimensions · mean only
Persuasion Index
A theory-guided framework that makes persuasive language measurable without collapsing analysis into one opaque model judgment.
My role: Lead computational developer and co-author, building evaluation pipelines, the web interface, REST API, CLI, and Python package.
Illustration above. Open the interface to explore the working analysis system.
ROOM / 03Spatial voice
Spatial Talk
A browser-based 3D meeting environment that uses spatial navigation and peer-to-peer voice to explore how virtual rooms can shape conversation.
My role: Independent designer and developer across concept, interaction design, 3D front end, real-time communication, and deployment.
Illustration above. Open the case study for the working 3D environment.
CLAIM
IMPACT
INSTALLATION / 2023Reveal what the label hides
Shadow Behind
An interactive installation using computer vision and physical participation to examine the distance between fast-fashion sustainability claims and their hidden consequences.
My role: Interaction designer and fabricator across concept development, visual reveal, physical installation, and computer-vision interaction.
Illustration above. Open the documentation to see the installation in use.
PARIS / 177513 linked lives
Slaves in Paris
A public-facing StoryMap that traces 13 enslaved people through eighteenth-century Paris and reconnects archival biographies to a georeferenced 1775 city map.
My role: Geospatial story designer and developer, integrating archival research, narrative sequencing, custom web work, and deployment.
Illustration above. Open the project to see the published StoryMap.
02 / Research
Systems that support interpretation.
I work from research question to operational definition, data pipeline, interface, and evaluation.
Research position
Patterns matter, but so do the assumptions and source evidence that produce them.
My research treats visualization as part of the analytical method. Interfaces should make aggregation useful without hiding uncertainty, provenance, or the path back to individual cases.
Peer-reviewed publications
2026Accepted · Main Conference
Persuasion Index: A Theory-Guided Framework for Persuasion Analysis
Liancheng Gong, Zhiyang Wang, Yiwei Xu, Julia Mendelsohn
EMNLP 2026
2025First author
Multi-Modal LLM Assisted Visualization of Large Visualization Collections
Zhiyang Wang, Devin Bayly, Carolina Roe-Raymond
PEARC ’25 · ACM
03 / Teaching
Ideas become clearer when students can test them.
I teach technical production as a way to ask sharper questions about media, interaction, and audiences.
Creative coding and making
Code! 1 & 2 · Interaction Lab
JavaScript, p5.js, Arduino, sensors, fabrication, and 3D printing, taught through small studies, individual debugging, and iterative project work.
Project development
Capstone · Applications
Research questions, prototyping, critique, collaborative practice, presentation, and portfolio development for undergraduate and graduate students.
Critical and networked media
Shared Minds
Machine learning, databases, and shared 3D environments examined alongside the cultural and communicative assumptions built into them.
04 / Contact