
An open-source port of Codex's Record & Replay capability — letting any agent learn and replay macOS desktop workflows from a single user demonstration.
I've built projects across content platforms, AI products, and full-stack web apps, including hands-on work with LLM, Agent, and RAG systems.

An open-source port of Codex's Record & Replay capability — letting any agent learn and replay macOS desktop workflows from a single user demonstration.
A consulting-grade presentation AI agent that turns ideas into structured, data-driven decks.

A quick selection of my latest writing and field notes.

Field notes and patterns for prompting agentic systems effectively, drawn from Claude Code and my own production experience.

Why context, not just model size, determines whether an LLM product holds up in production.

Why Model Context Protocol matters: a practical explanation of how MCP standardizes access between LLMs, tools, and external systems.

My reading notes on Nexus and why AI may become the first information technology capable of becoming a node inside civilization itself.