Tracing AI Handwriting Breakdowns Across the System
Mapped handwriting issues to model, sensor, protocol, or interaction layers.
Turned ambiguous failures into routed QA criteria.
Alex Rudnicki
Research and strategy for complex human systems.
Selected Work
View all workReframing Slide Access as a Live Workflow Problem
Mapped live slide-access gaps for Blind and low-vision participants.
Defined live-access criteria for slide delivery.
Scaling Research Infrastructure for AI Accessibility Studies
Built a verified Deaf and Hard-of-Hearing recruitment and access workflow.
Converted access logistics into a reusable study workflow.
Research Highlights
A generic non-invasive neuromotor interface for human-computer interaction
Contributed to large-scale Reality Labs studies on handwriting and gesture models designed to generalize across people.
Opens in a new tab.Making Social Video Accessible by Design
Analyzed 1,000+ social videos for NSF-funded research identifying recurring access barriers across digital video platforms.
Opens in a new tab.About Me
I thrive in complex, multidisciplinary environments, using mixed-methods research to make systems and experiences more understandable, accessible, and human.
I’m a researcher with a background in psychology, HCI, accessibility, and human-AI interaction. My work starts where the problem is still unclear: a model issue, a design gap, a workflow gap, or an assumption no one has questioned yet.
My background in ASL and Deaf studies shaped how I think about access: not as accommodation, but as infrastructure. Inclusion belongs in systems from the start, not added after the decisions are made.
Perspectives
-
Accessibility is infrastructure
Systems work better when disabled people help shape them from the start. Accessibility is not edge-case support or compliance. It is infrastructure for systems people can understand, trust, and use.
-
Models are not the whole experience
Benchmarks show what a model can do. Research shows whether people understand it, recover from failure, and trust it in use. Evaluation has to measure the relationship, not just the output.
-
The whole system matters
When parts of a system do not connect, the burden falls on people. They repeat information, catch errors, and fill in what was missed. Systems work best when information carries through.
Contact
Building something complex?
Open to collaborations and conversations around complex systems, accessibility, and human-AI interaction.
PDF document, 2 pages. Downloads directly.