Research
What I work on
I look for places where technology rests on assumptions about disabled people that do not hold, and I design better options together with the people affected.
AI that leaves the choice to you
AI can save effort by deciding things for you, and every one of those decisions is a guess about what you want. The tools I build split a task into small, clear choices, so the model does the heavy lifting and the person stays the author.
StepWrite drafts longer messages by voice, and Anteroom applies the same idea to working through messages. Virtual Buddy and PromptAssist cut the typing needed to set up a chatbot or to write an image prompt. I also study what happens when AI advises and remembers: an experiment on AI coaching with 160 adults, and interviews about what assistants remember about disability.
Papers in this area
Interaction that feels like yours
Keyboards, mice and game controllers are built for two agile hands, and walking in virtual reality is designed around people who walk. Alternatives usually aim only at getting the task done. I start from what a person’s body can do, and I treat how the interaction feels as part of what has to work: the touch of a click, a sense of control, the experience of moving.
MouseClicker presses a real mouse for you and lets you feel the click. A camera that reads facial expressions became a game controller, tested by players with severe motor impairments. Virtual Steps explored what walking in virtual reality is like for someone who has never walked.
Research from lived experience
My own experience as a disabled person tells me where to look. The answers come from working with other disabled people, who are the experts on their own lives.
An experience report at ASSETS argues that this kind of insider knowledge is rigorous. Co-design runs through my projects: an image-generation tool shaped by a team in which most of us have motor disabilities, and movement in virtual reality designed with people with mobility disabilities.