Foundations

Math and ML foundations for practical builders

The useful version of ML math is not memorizing symbols. It is knowing what a model is optimizing, what the data says, and how to tell when the result is lying to you.

By Sean Findley May 3, 2026 5 min read

Useful foundations

The practical foundation for machine learning is understanding representation, optimization, uncertainty, and evaluation. The details matter, but they matter most when tied to a working system.

A builder's map

  • Vectors describe how inputs become something a model can compare and transform.
  • Loss describes what the model is being asked to improve.
  • Gradients describe how the model receives direction during training.
  • Evaluation describes whether improvement on paper is actually useful in the product.

Keep the math connected to behavior

Mathematical intuition becomes powerful when it helps explain model behavior, not when it turns into decoration. The point is better judgment: what changed, why it changed, and whether the result should be trusted.

Need this kind of engineering?

Polished systems, practical AI, and product work that survives production.

I am available for senior full-stack engineering, AI product engineering, and modernization work where clarity, shipping discipline, and tasteful UX matter.

View resume Start a conversation