Agentic engineer. Founder of Evidence Code Labs.
I build AI-native systems that compress quarters into weekends. Daily driver: Arch + Claude Code + Jeff — my Personal AI Infrastructure that remembers what works and gets sharper every session.
I'm not a vibe coder. I graduated from that. The bet I'm making: context engineering + agentic workflows + a real papertrail of trust beats throwing prompts at a model and hoping.
ContentEngine — Agentic content pipeline. Arbitrary-input ingestion, adaptive triage, rejection rubrics. Designed as a CLI/MCP for technical operators, not a UI product.
sales-rpg-ai — Real-time AI sales coach. Listens to calls, surfaces objection responses live. I built it for myself — close rate went 5% → 25% in a month, 25% → 30% the next month, eventually then 40%.
rough-remotion — Animated hand-drawn diagrams rendered to MP4. Remotion + rough.js + React + TypeScript. For content creators who hate generic explainer aesthetics.
cognite-industrial-sensor-monitoring — A coding-challenge solution showing how I work end-to-end with an AI-native loop. Same delivery quality I bring to client work.
Personal_AI_Infrastructure — My fork of Daniel Miessler's PAI, customized for how I actually use Claude Code every day.
kubecraft-yt-transcript-api — Small but useful: pulls YouTube transcripts as an API.
- Everyone's Talking About AI Loops. They're Missing the Point.
- How I Graduated from Spec-Based Vibe Coding to Agentic Engineering
- Clear Thinking is the Top 1% Skill in the Age of AI
- The AI-Augmented Learning Thesis
- Semantic Blueprints: Separating Domain Knowledge from AI Orchestration
- Blog: theaustinjohnson.com
- Twitter / X: @eccentric_node
- Hire me through Evidence Code Labs