Engineering, AI, and civic-tech projects I've built or led — from test frameworks at Adobe to a multi-agent chaos lab of my own.
A tiny, cheap, fast multi-agent chaos lab. Drop a different LLM into every villager and watch a society emerge — economy, governance, relationships, all instrumented. Backed by a 1,284-test suite across 200+ commits.
A curated collection of Claude Code skills and agents — the methodology behind a multi-agent development ecosystem. Orchestrator patterns, reusable skill composition, operational contracts. Also the studio mark in the footer.
Performance and integration testing of a distributed LLM orchestration platform — gateway-vs-direct latency, MCP proxy, multi-provider. A reusable NeoLoad practice turned "feels slow" into a number. (Anonymized within NDA limits.)
A tool-agnostic specification for the SKILL.md format — so a skill written once is portable across agent runtimes. A small, sharp standard underpinning the broader Skill Madness ecosystem.
Assembled and led the team that revamped Magento's functional automation framework, prototyping the UI testing approach that became the foundation for its functional test tooling — still referenced in Magento's lineage.
A functional-UI demonstration — the FUI research behind the sci-fi terminal aesthetic, built to actually run as a live, typeable terminal. The aesthetic seed for earlier versions of this site.
A civic-engineering tool surfacing congressional fact sheets for legislative advocacy — clean, fast, factual reference material in front of the right people. Built with Vue 3 and .NET.