I studied mathematics at Reed College and then spent roughly a decade as a long/short equity analyst and principal investor. The mandate was almost always the same: work across sectors, understand complex systems deeply enough to see around corners, and express that view in risk-adjusted positions.

Starting around 2010 I got pulled into deep learning, back when handwritten digit datasets and Restricted Boltzmann Machines were still a big deal. That background ended up being unusually helpful: you can’t really evaluate the claims around AI infrastructure without understanding, at least qualitatively, how the models actually work.

Over time those threads converged. Lumera Network is my attempt to build infrastructure where storage, compute, and interoperability are open and verifiable. SmartEdgar and my other tools are designed for a world in which analysts have fleets of agents working alongside them, not just spreadsheets and search engines.

The through line is simple: take complicated systems, understand them at a mechanistic level, and then build things that make that understanding actionable for other people.