AI Journey
The public log of a transition from systems programming to AI — math, ML fundamentals and end-to-end projects, with one rule: never train a model I can’t explain.
Code ↗Context
A structured roadmap from foundations to research: advanced Python and algorithms, the math for AI (linear algebra, calculus, probability, optimization), classical machine learning, then deep learning with PyTorch — each stage with notes, exercises and implementations, logged day by day.
Current focus: Trail, a portfolio bot. The V1 intent classifier is trained by hand — TF-IDF + logistic regression, F1 macro 0.629 on 5-fold cross-validation — and is being integrated behind a FastAPI endpoint with a confidence threshold and a response policy.
House rules: clean, documented code; every project ships with results and conclusions; fundamentals over hype.