❯ tail -f ~/writing/*.md ▌
Writing on Agentic AI, GenAI, AI Security & Open Source
// Essays on Agentic AI, GenAI, AI sovereignty, GNU/Linux and open-source security — written from two decades of practitioner experience.
I write long-form on the things I'm building and arguing for: Agentic AI frameworks, GenAI in production, the case for AI sovereignty, and why GNU/Linux and open source remain the right substrate for ethical, secure enterprise growth. Pieces lean practitioner-first — patterns, trade-offs, and the lessons that only show up after you've shipped.
// latest on substack
Eight Bits of Trust: The eMudhra CA Crisis and the Number 4 ↗
How an intermediate private key literally equal to "4" went undetected for 17 months, forcing emergency security releases across browsers, Go, Python, and global edge proxies.
Generating Code is Solved. Shipping Reliable Software Isn't. ↗
Generating code fast is a solved problem. Any team with an API key can pump out features at five times the volume they used to.
Open Weights Are Not Open Source ↗
Everyone says "open source AI." Almost nobody ships it. Here's what the licence actually says, and why your legal team will care before your engineers do.
The "Throat to Choke" Economy: Why AI Won't Take Your Job Title ↗
Silicon is absorbing the labor of execution while privatizing the liability to carbon. Welcome to the era of the designated human scapegoat.
The Orchestration Layer Is the Only Thing in Your AI Stack That Should Outlive 2026 ↗
On durable execution, the framework trap, and why background agents will break most production setups before they help them
The Hidden Engineering Work Behind Enterprise Agents ↗
The industry is dramatically underestimating the work it will take to put agents into production inside large organizations.
More in the full archive ↗
Subscribe on Substack for new essays, or follow me on Medium for archived pieces.