QUASI CONVEX UNION

Engineering, design,
and research.

Projects.

Systems & protocols, and the ideas behind them.

Selected projects

01 / Protocol Halliday Agentic computation across blockchains, constrained by user intent. 02 / Protocol Compound Lending, liquidation, and decentralized governance. 03 / Protocol & game theory Ether Freakers Programmable ownership, pooled Ether, and emergent strategies. 04 / Real-time pricing AdRoll Real-time pricing, compact analytics, and distributed computation at advertising scale. 05 / Storage & indexingJFDBAtomic indexing and prefix search in a compact embedded database. 06 / Product & interaction Skej Agents negotiate meeting times and places while preserving privacy. 07 / Talent intelligence Chive Learning who to approach, and when, from changing signals and expert judgment. 08 / Machine learning Runway 20 An assistant connecting email, calendars, and location to timely action. 09 / Distributed computing Disco From scientific computing to tools for discovery at scale. 10 / Data visualizationvziProgrammable visual exploration, from Unix streams to the browser.

And much more.

These projects offer a few ways into a much larger body of work. Long before the financial protocols came scientific computing: processing microarray experiments and early next-generation sequencing data, building tools to investigate questions at the scale of the genome. Work on distributed computation, peer-to-peer live video streaming, and advertising systems brought a different intensity, with enormous volumes of data, tight response times, and decisions whose consequences returned through the next round of feedback.

Along the way, the tools themselves became subjects of exploration. A visualization could make an unfamiliar pattern visible; a small language or library could make a difficult idea easier to express. Programming language design and formal verification opened further possibilities for describing systems precisely and reasoning about their behavior. The work in markets extended those questions to incentives and game theory, where the behavior of the participants becomes part of the system being designed.

There is room here for mathematical curiosity as well as immediate application. Information theory, stochastic optimization, topology, and knot theory belong to that wider exploration. Some ideas find their way into working software; others remain questions worth pursuing. A useful connection may emerge years later, in a setting that was nowhere in view when the work began.

The next project need not resemble any of these. It might begin with a research question, an ambitious product, a system that has outgrown its original design, or an intuition that something should be possible. There is space to follow it from the first experiment into the architecture, mathematics, and fine details that make it real. This collection is an invitation to imagine what that work could become in your world.