Calculus
How models learn. Every weight update, every backprop pass, every optimizer is calculus applied to a loss function — from first principles.
derivatives
partial derivatives
chain rule
gradient
gradient descent
backpropagation
convexity
Topic overview
Start with derivatives as local change, then connect partial derivatives, gradients, gradient descent, backpropagation, and convexity into one optimization story.
Core concepts
Know slopes, chain rule, multivariable derivatives, gradient vectors, learning rate behavior, loss surfaces, and why backprop is repeated chain rule.
Why it matters
Calculus explains how models improve: every parameter update depends on estimating which direction reduces error.
Interview relevance
Calculus tests whether you can reason about optimization, vanishing gradients, loss behavior, and training stability beyond memorized formulas.