Building agentic AI systems that reason, act, and self-correct.
Three projects. Each one an architecture narrative exploring design decisions, tradeoffs, and the systems thinking behind autonomous AI agents in production.
shopify-agent
Multi-agent orchestration with tiered model selection and auto-fix loops
Placeholder — card copy to be provided. Architecture explores model-tier cost optimization, DOM-to-Liquid source mapping for automated accessibility fixes, and defense-in-depth permission controls across CLI flags, runtime callbacks, and hook layers.
Mission Control
Voice-powered desktop agent with MCP orchestration and ambient monitoring
Placeholder — card copy to be provided. Architecture explores regex-over-LLM intent classification, project-scoped MCP server isolation, ambient health monitoring with severity-filtered alerts, and mock/live toggling for credential-free development.
shopify-docs
Zero-dependency MCP documentation router with intelligent query routing
Placeholder — card copy to be provided. Architecture explores keyword-length scoring over embeddings, hand-implemented JSON-RPC 2.0 over stdio, pre-resolved library IDs eliminating network round-trips, and the engineering case for zero dependencies.