The AI Career Constellation

The source describes seven stages and eight directed prerequisite edges. Foundations leads to how AI thinks, which branches toward transformers and tokens. Tokens also leads to transformers and retrieval augmented generation. Transformers leads to agents. The retrieval and agent branches join at deployment. This two dimensional diagram represents those source edges, rather than a projection of the live three dimensional scene. The prerequisite lists explain relationships but do not prevent users from selecting future stages. The progress slider stores a number from zero to seven. Nodes with an index below that number are styled completed, and the percentage is that count divided by seven, rounded. At progress three, nodes zero, one, and two are lit and the label is forty three percent. Node two lists node three as a prerequisite, but node three is still unlit. This demonstrates that completion follows a numeric prefix, not validation of the graph or proof of skill mastery. The page starts at progress one, so the displayed question uses index one, about backpropagation. A correct choice adds one to progress, capped at seven; a wrong choice keeps the same progress. The question function then renders the next current question. Moving the slider directly to seven marks the constellation complete without answering any quiz. These checks are study aids, not assessment evidence, credentials, or proof that a learner can deploy an AI system. Selecting a star enlarges its sphere by one point five and fills an inspector with its stage text and prerequisite names. Each graph edge lights only when both endpoint indices are below progress. There is no durable progress storage in this script; a fresh load begins at one again. The live drawing needs external Three J S, which is blocked in our offline checks. Consequently our preserved static controls and resized layout do not verify actual WebGL buffers, orbit interaction, node picking, or quiz execution.