AI-Powered Search: Modern Retrieval for Humans & Agents

Understand emerging Agentic Search strategies, implement them with the Agentic Search loop, and be at the forefront of this paradigm shift.

Optimize search for LLMs and agents, as well as the critical strategies (chunking, adaptive queries, guardrails, etc.)

RAG represents the largest growing use case for search, and "retrieval" is by far the hardest part of RAG.

Learn how to integrate your content's context, your domain context, and your users' individual contexts to optimally interpret query intent.

We'll cover query classification, query-sense disambiguation, and semantic query parsing, using multiple lexical and LLM-based strategies.

Dig into emerging techniques like Wormhole Vectors, Sparse models (Semantic Knowledge Graphs & SPLADE), & System One / Decision Models.

The field of information retrieval is vast. Learn the mental models that tie it all together.

Understand when to use bi-encoders vs. cross-encoders vs. knowledge graphs, and ANN vs. quantization vs. representation learning.

Understand multimodal and hybrid search, search on semi-structured embeddings, and how to properly mix sparse + dense retrieval techniques.

Leverage "reflected intelligence" from your use behavioral signals to automatically improve your ranking models.

Learn to implement algorithms for signals boosting, matrix factorization, and knowledge graph learning

Implement personalized search by building real-time user behavioral embeddings leveraging matrix factorization / collaborative filtering.

Learn common Machine-Learned Ranking (Learning to Rank) models put into production, from LambdaMART to cross-encoders

Get hands on experience performing feature engineering and representation learning to optimizing AI Search for production.

Implement ANN and quantization strategies, reranking, semantic caching, and even local model serving for search and RAG.

Learn how to systematically develop great ranking features using a combination of vector and lexical search indices.

Generate automatic judgments from user click stream data using Click Models

Overcoming ranking biases (presentation, position, & confidence bias) + uncover unexplored ranking features using SDBNs and Active Learning.