Writing
Notes on AI engineering, computer vision and building products.
From the OpenAI - Hugging Face incident to who actually owns secure software: AI is not magic, but its speed and scale have changed the game.
Office politics does not have to mean scheming or rivalry. Sometimes it begins with a coffee, a lunchtime game, and the trust we build through small acts every day.
CIA, AAA, vulnerabilities, threats, exposure, controls, and four risk responses - a practical foundation for moving from finding bugs to reducing business risk.
Social good is not only big AI-for-good initiatives - it is the small, slightly costly choices we make at work every day for the benefit of the many.
A simple mantra that makes project meetings more open and honest - so a team can stop defending and start improving the product together.
How I use Microsoft’s AI Engineer Coach - a local-first VS Code extension that turns your AI session logs into a dashboard - to keep my AI-assisted coding honest.
A short take on Han-chung Lee’s essay: hand the routine work to AI, but keep the understanding and the verification for yourself.
A short take on Han-chung Lee’s essay: why top-down "everyone must do AI" mandates echo China’s Great Leap Forward - and produce activity instead of outcomes.
How we combine OCR, machine translation, inpainting and LangChain multi-agent orchestration to translate comics entirely on local GPUs.
Contact
Open to AI engineering roles and collaborations. Reach out anytime.