/// Project Dossier

Neural Network from Scratch

A ground-up neural network implementation in pure Python — no frameworks, just math. Covers neurons, layers, activation functions, and loss calculation.

Focus

Deep understanding of neural network fundamentals by implementing every component from first principles.

Milestone 01

Built forward-pass computation with dot products and layer objects

Milestone 02

Implemented ReLU and Softmax activation functions from scratch

Milestone 03

Coded categorical cross-entropy loss calculation

Milestone 04

Demonstrated multi-input neuron and multi-layer architectures