Machine Learning Starts Before the Model
In real systems, the hardest part of machine learning is often not the algorithm. It is the work of making data coherent, reliable, and usable enough for modeling to matter.
Notes on systems, machine learning, and technical judgment.
Writing is where I refine how I think: about complexity, clarity, products, and the craft of technical work.
Scientific Reasoning for Applied AI
A scientific background is useful in applied AI not because it makes the work abstract, but because it trains you to question assumptions, isolate variables, and take evidence seriously.