Posts

• Programming Languages

  Why Lisp Seems Easy for LLMs

  March 16, 2026

  LLMs often look unusually good at small Lisp programs because the code already exposes most of its tree structure.

• Systems Engineering

  Luna: My Trusted, Autonomous AI Agent

  February 13, 2026

  A supervised autonomous agent that runs work loops with cryptographic approvals, strict guardrails, and a clear human authority chain.

Preprints

• Preprint

  Phased Autonomy: Infrastructure for Self-Directed AI Agents

  February 2026

  Complete architecture for escalating agent autonomy through validated safety infrastructure, structured memory, and hardware-backed operator authentication.

• Preprint

  Cryptographic Operator Authentication for AI Agents

  February 2026

  Hardware-backed ECDSA protocol for verifying the human in the loop, with threat model, offline verification, and continuous physical presence.