Notes on data, AI, IT
and security
No marketing fog. The way I think about real problems with founders and managers.
Legacy on the factory floor: securing Windows XP and 7 systems you cannot patch
Industrial equipment lives 15 to 30 years, and the control PCs under it run on Windows XP and 7 that lost support long ago. A routine patch can void the vendor warranty or crash the process. What to do when you cannot update: move the defense from the host to isolation and compensating controls.
AI on the factory floor and a new class of attack: how to protect an industrial model
Factories put neural nets on predictive maintenance and process optimization. A new class of attack - poisoning the training data and quietly spoofing telemetry - pushes the AI into decisions that wreck equipment. Where the real line of defense actually runs.
"Do we have to keep expensive data scientists on payroll forever now?" What you actually need after delivery
The owner's fear at the end of a data project: that expensive specialists now have to stay on staff forever. What is actually expensive, and two honest ways to hand the work over without bloating payroll or staying locked to the contractor.
"We have a niche stack - have you worked with it?" How I take on unfamiliar technology
A direct answer to the client's real fear: paying my rate while my engineers google your technology. What I actually sell, and how I ramp on an unfamiliar or legacy system.
Local AI: running a model so your data never leaves the perimeter
An honest engineering and cost picture of local AI: what your own perimeter actually buys you, the ladder of options from Ollama to air-gapped, and when a hosted business tier is enough.
Can you delete your data from an already-trained AI model
Why removing yourself from a trained model is nothing like deleting a database row, what actually sits behind unlearning and the right to erasure, and where the real leverage is.
Can an AI hand my personal data to another user
An honest split: where a model can actually recall someone else's data from training, and where the real leak comes from your own system - RAG, logs, and access control.
How to stop AI from training on your prompts and chats
An honest walkthrough for an individual and for a company: where chatbots train on your conversations by default, how to switch it off, and why opt-out never erases what a model already learned.
Experience in IT doesn't appreciate on its own
The market pays not for years in the profession but for the ability to solve current problems. Why expertise quietly loses value, and which part of it barely depreciates at all.
When AI exposes the debt sitting in your codebase
The first numbers from Anthropic's Glasswing project are not a story about a smart model. They are a story about how much old vulnerability lives in code we use every day.
Data contracts: from principle to working tooling
Data contracts were discussed as a concept for several years. In 2026 they are working tooling with real implementation costs and real results.
Robotics after LLM: why the next question is not chat, but action in the world
LLMs changed the interface for interacting with computers. The next shift is physical systems that understand context and act in the real world.