Leon Zhu

I am an independent developer working on AI development infrastructure, distributed systems, protocol design, and applied quantitative finance. My work spans long-horizon agent workflows, consensus-level infrastructure, and trading systems, with a focus on building things that are correct, verifiable, and grounded in real constraints rather than abstraction.

My work includes mandrel, an open-source protocol for running AI coding agents on the same codebase for months; QuantX, a Go-based quantitative trading platform; and the technical architecture of HKChain, a compliance-native blockchain for stablecoin issuance and real-world assets.

Across them, I am interested in how systems preserve correctness and intent over time: how an AI coding session inherits decisions without inheriting an unbounded transcript, how compliance can be enforced at the infrastructure layer, and how trading ideas survive contact with real execution. If that overlaps with what you are working on, I am reachable below.