Hello, I'm Armon. I'm a self-taught developer learning full-stack development with TypeScript, Next.js, and Node.js, as well as backend systems including MCP and distributed architectures. I graduated from Dartmouth with a degree in quantitative economics.

I'm particularly interested in AI safety through evaluation and red-teaming approaches. I'm currently building AgentRedChain to explore whether sparse evaluation methods from psychometrics can reduce the cost of security testing multi-agent systems. I'm particularly interested in how attacks propagate through agent networks, how you test for adversarial robustness when the system itself is distributed, and how you assign liability when failures emerge from interactions rather than individual components. These are empirical questions that need systematic testing infrastructure. The work is still early and unvalidated, but I'm learning through building.

You can find some of my projects and experiences below. In my free time, I like to buy and fix old jeeps, draw with charcoal, and go skinning uphill.