About

I came into technical work through music. I studied performance and theory at the University of North Texas before moving into computer science at Rochester Institute of Technology. My graduate work put the two together, through machine learning, music generation, and scientific sonification, and produced published research on transformer-based music generation and the sonification of black-hole merger data.

Since then the work has broadened considerably. I've built real-time detection systems running at the edge, distributed training infrastructure for air-gapped environments, storage and incentive protocols operating in production, retrieval and agent systems, and a multi-tenant AI product carried from initial product thesis through deployment and day-to-day operations.

I'm most useful on problems that don't fit neatly inside one layer of the stack. Sometimes the difficult part is the model. Sometimes it's the protocol, the deployment environment, the data architecture, the release process, the incentive structure, or the product boundary. Often it's the interaction among several of them. That breadth is deliberate, though I don't think of myself as a generalist for its own sake — I like understanding a system well enough to identify what is actually load-bearing, and then following the problem wherever that leads.

I'm also interested in the physical systems underneath ordinary life: power, water, heat, shelter, tools, repair, and the practical knowledge required to maintain them. Software has a way of making abstraction feel self-sufficient; physical systems are a useful reminder that eventually something has to work in a particular place, on a particular budget. Capability Commons grew out of that — good information is useful, but usable capability is better.