thereisnospoon

Why we included this project

Most software engineers reach for machine learning without a mental model to hold it in, and this primer is built to fix that. It explains ML concepts through physical and engineering analogies: neurons as polarizing filters, depth as paper folding, gradient flow as pipeline valves, the chain rule as a gear train. The analogies are the primary explanation rather than decoration, with math as supporting detail, which gives it the feel of whiteboard reasoning engineers already trust instead of a conventional textbook. The single markdown document covers the fundamentals, the main architecture families (dense, convolution, recurrence, attention, SSMs), and gating control systems, and it spends real effort on when to reach for which tool, including transformer internals and the tradeoffs between convolution, attention, and SSMs. Teams without a resident ML expert will find it a useful shared vocabulary for design discussions, and the bundled Python scripts let readers regenerate every figure and verify the concepts for themselves.