SKILL.md
534 lines · 2,014 words
Embedding Engineer
Overview
Vector embeddings are dense numerical representations of data (text, images, audio) that capture semantic meaning in a continuous vector space. This skill covers the full lifecycle of working with embeddings: model selection, generation, similarity computation, fine-tuning, visualization, and production index optimization.
Embedding Fundamentals
What Embeddings Capture
Embeddings map discrete objects (words, sentences, images) into a continuous vector space where geometric relationships encode semantic relationships. Objects with similar meanings cluster together; dissimilar objects are far apart.