Offered in partnership with Dell Technologies. This session introduces embedding models as a core component of modern machine learning systems for clustering, ranking, and large-scale information retrieval. It explains how embeddings are trained, covering representation learning theory, contrastive and metric-learning losses, and data preparation strategies that influence embedding quality. An EmbeddingGemma‑style code walkthrough illustrates dataset construction, pairing, and negative sampling, and connects training choices to retrieval performance using a real-world agentic search application at Dell.

Level of Difficulty: Beginner to Intermediate