There is No Spoon. A software engineers primer for demystified ML

A machine learning primer built from first principles. Written for engineers who want to reason about ML systems the way they reason about software systems.

You're a strong engineer. You can draw a software system on a whiteboard from your own hard-earned mental model. You understand tradeoffs — maintenance vs elegance, performance vs complexity.

You have a gut for software design. You don't have that gut for machine learning yet.

You know the tools exist but you can't feel when to reach for which. This primer builds that intuition.

This isn't a textbook or a tutorial. It's a mental model — the abstractions you need to reason about ML systems the way you already reason about software systems.

Every concept is anchored in physical and engineering analogies:

• Neurons as polarizing filters
• Depth as paper folding
• Gradient flow as pipeline valves
• The chain rule as a gear train
• Projections as shadows

These analogies aren't decorative — they're the primary explanation, with math as the supporting detail.

The focus is when to reach for which tool and why — not just what each tool does, but the design decision it represents and the tradeoffs it implies.

The primer is organized in three parts:

Part 1 — Fundamentals

The neuron, composition (depth and width as paper folding), learning as optimization (derivatives, chain rule, backprop), generalization, and representation (features as directions, superposition).

Part 2 — Architectures

The combination rule family (dense, convolution, recurrence, attention, graph ops, SSMs), the transformer in depth (self-attention, FFN as volumetric lookup, residual connections), encoding, learning rules beyond backprop, training frameworks (supervised, self-supervised, RL, GANs, diffusion), and matching topology to problem.

Part 3 — Gates as Control Systems

Gate primitives (scalar, vector, matrix), soft logic composition, branching and routing, recursion within a forward pass, and the geometric math toolbox (projection, masking, rotation, interpolation).

Interactive exploration with an AI agent

This is the more powerful approach, and closer to how the primer was actually built. Feed the primer (or a section of it) to your preferred AI coding assistant and explore it conversationally. Ask "why" questions. Propose wrong answers and see if the agent catches them. Ask for concrete examples. Ask what would happen if you changed one thing. Ask how two concepts relate.

The primer gives both you and the agent a shared vocabulary and a correct conceptual framework — the conversation fills in everything a static document can't.

The primer is the map. The conversation is the territory.