Embedding map
Every SOUL embedded with a sentence-transformer model, then projected into two dimensions. Minds that think alike land near each other. Switch the model and the projection, drag to pan, scroll to zoom, search to highlight, and click any point to open it.
Technology Healthcare Skilled Trades Engineering Science Education Business Government Military Law Transportation Agriculture Hospitality Creative Sports Entertainment Finance Public Service Historical Emerging Life Roles
Projections are precomputed (scikit-learn + umap-learn) so the map is instant. PCA is linear; UMAP and t-SNE are non-linear — in those, read clusters and neighbourhoods, not absolute distances. Tuned defaults: UMAP n_neighbors=15 / min_dist=0.1, t-SNE perplexity=30, cosine metric.