From intuition, to math, to code.
I’m Irene Markelic — a computer scientist (Ph.D.) working in applied machine learning. This site is where I break down the ideas behind AI and ML properly: not just the formula, and not just “here’s the code” — but why it works, step by step, until it actually makes sense.
Every piece here follows the same path: start with the intuition, build up the math without skipping steps, then ground it in code.
A note on where things stand: I’m currently restructuring this site to make it easier to navigate and to better reflect where I want to take it. Some pages are being reworked over the next few weeks — thanks for your patience while I get everything in order.
New here? Start with a suggested path
If you want to build real intuition for machine learning from the ground up rather than jumping in at a random article, here’s a suggested order. Each track builds on the last.
Track 1 — Foundations you’ll keep coming back to
The building blocks that show up again and again once you know to look for them.
- Understanding the Geometry of Orthogonal and Diagonal Matrices
- Understanding the Standard Normal Distribution Formula
- What is a Logit?
Track 2 — Core tools of machine learning
The classical methods that most of modern ML is still built on top of.
- Linear Regression: Statistical vs Machine Learning View
- A Brain-Friendly Guide to PCA: Math, Visuals, Code
- Deriving the Singular Value Decomposition (SVD) from First Principles
Track 3 — Deep learning and modern architectures
Where it all comes together — the mechanics behind today’s models.
(More tracks — probability & statistics, optimization, and a full transformer series — are on their way.)
Or browse by topic
Stay in the loop
Follow along on LinkedIn or Medium for new posts as they go up.