Machine Learning Enthusiast
Building machine learning systems from first principles.
> Building TerraSeg — Satellite Image Segmentation
ENTRY — PERSONAL LOG
I approach machine learning the way an engineer approaches a bridge: not as a black box to invoke, but as a structure whose every beam and joint should be understood before it is trusted. Most of my work begins by tearing a well-known method down to arithmetic — a loss function, a derivative, a loop — and rebuilding it by hand until the magic disappears and only mechanism remains.
This site is kept the way a lab notebook is kept: dated, numbered, occasionally messy, always honest about what is finished and what is still on the bench.
A growing collection of systems built from first principles — including my ML-from-scratch work, where I implemented seven machine learning algorithms without relying on ready-made model implementations.
A machine learning library built from first principles, implementing seven core algorithms without relying on ready-made model implementations. The project focuses on understanding the mathematics, optimization, and mechanics behind the abstractions hidden inside `.fit()` and `.predict()`.
Field notes and long-form writing — the reasoning behind the code, kept for whoever reads the notebook after me.
Building TerraSeg — Satellite Image Segmentation
LAST UPDATED · 2026-09-06Open to research collaboration, engineering roles, and honest conversations about how things actually work.
