Work

My work focuses on the layers between computation and human sensemaking, where interpretation, cognition, and power get encoded. I combine research, design, and implementation to explore these questions in situ.

I come from HCI and interface design — building systems where tiny interaction choices decide what people notice, remember, and believe. Over the last decade, the questions I explore have expanded. I now work on what I call epistemic infrastructures: the technological architectures that support people in staying oriented as AI increasingly mediates cognition.

Substrates that resist collapse

Most systems optimize for convergence — the answer, the summary, the first plausible output. This work explores systems that resist premature closure.

When retrieval and generation systems optimize only for relevance and fluency, they can repeatedly reinforce familiar associations. Exploration narrows. Alternatives disappear. I build technology that creates productive friction: preserving alternative interpretations, surfacing contradictions, and letting structure develop through interaction.

Agency and coordination in distributed cognition

As AI systems mediate how we find, interpret, and remember information, they also shape the conditions under which we exercise agency. Who controls what shapes your attention? Can you trace how a conclusion was reached? Can your accumulated knowledge move between tools? Can you decide which parts are shared?

Agency is also relational. People and agents need to cooperate, coordinate their actions, and negotiate shared contexts while retaining control over their own knowledge. This work explores provenance, context portability, and selective exchange as conditions for that cooperation. membranes.network develops this question at the protocol level: how independently governed systems can exchange context while retaining control over their boundaries.

Knowledge as territory

Knowledge as territory, not inventory. Exploration opens unfamiliar paths. Exploitation makes use of established knowledge. Spatial interfaces should support both. These projects explore how spatial thinking and embodied cognition can make high-dimensional information landscapes navigable. Position, distance, scale, and recognizable landmarks give us ways to organize relationships and return to places we’ve encountered before.

I develop interfaces that let people move between an overview and individual sources without losing orientation, using terrain metaphors and visual variance to make complex structures easier to explore. Maps, not folders. Territories, not hierarchies.

Design as meta-discipline

Interfaces are becoming temporary cognitive surfaces. Assembled for a task, reshaped as the inquiry develops, dissolved when done.

This changes the role of design. We design the conditions under which interfaces emerge, including the structures and constraints that keep them coherent. Design systems become grammar; interfaces become sentences generated for the situation. What persists is the substrate: sources, relationships, provenance. The challenge is maintaining coherence and continuity as experiences and their interfaces reshape around intent.