QA. Software Engineering. Cloud. Transformation.
Wasn't a plan, just a career's worth of saying yes and figuring it out as I went.
This is where I write down what I learned. If it helps you, great. If not, it's been good therapy.
0:50 Amazon Bedrock gives a company over a hundred AI models from different makers through one Amazon account. What happens to your data, and the catch of letting Amazon run the models.
1:14 Sovereign AI is a country making sure it doesn't depend on a few foreign companies for its AI. What Britain is doing about it, why, and the catch that nobody builds it all alone.
3:30 Before an AI model can answer you, it reads a vast amount of text and sits one test, guessing the next word, trillions of times. What goes in, what gets thrown out, who marks the guesses, what the model's numbers are, what it costs, and why there's no text inside it at all.
2:55 A model that has only read the internet can continue text but can't yet help you. How people turn it into an assistant, by writing good answers, ranking its own, and training a second model to score them the way they would. Reinforcement learning from human feedback, in plain words, and whose taste it learns.
AI Compass
An adaptive thought piece on AI strategy, not a tutorial or a roadmap, but a set of principles to help you think clearly as the landscape shifts.
Read the compass →Learn
Plain-English breakdowns of how the technology actually works. Start with AI Fundamentals, a short video series on what's really happening inside an AI model.
Start learning →Blog
General tech stuff, career observations, and the occasional opinion I'm willing to put my name on.
Read the blog →Projects
What's going well, what fell off a cliff, and what I'm still convinced will work given enough weekends.
See the projects →Latest posts
All posts →
AI prototypes look finished, win the room, and set a bar the real system then has to clear. 'Build on it or start again?' is the wrong question. The code was never the expensive part. The promise was.
I built a pipeline that turns a text file into a narrated, subtitled training video in my own voice, running entirely on my Mac. The code was the easy part. What it taught me about quality is the part worth sharing.
A two-agent CrewAI pipeline that turns an article and my rough thoughts into a LinkedIn post in my voice, running entirely on local hardware. The build was straightforward. Teaching a local model to sound like me is the part I'm still working on.
Tell me what you're stuck on. I take requests, and the good ones end up as a video.