Exploring Modern AI Search
Today’s retrieval systems operate at massive scale (100 billion vectors) and are undergoing exciting AI-driven paradigm shifts to improve retrieval quality, including new approaches for Multivector, Multimodal, Hybrid, and Agentic Search. In this series, we’ll explore both the foundational principles of modern AI Search, as well as what’s being implemented at the frontier of information retrieval.
By continuing, you agree to Maven's Terms and Privacy Policy.
Thu Feb 26·7:00 PM UTC
Searching 100 billion vectors in object storage
In this Lightning Lesson, Nathan will introduce turbopuffer, a new search database built from first principles on object storage. He will introduce the architecture of the system at a high level before diving deep on how it was recently enhanced to serve production search at 100 billion vector scale.
You'll learn from
Nathan VanBenschoten
Chief Architect at turbopuffer
Doug Turnbull
Principal Search Engineer
Trey Grainger
Founder @ Searchkernel, Author "AI-Powered Search"
Fri Feb 27·4:00 PM UTC
Optimizing Search & Data Processing with Self-hosted SLMs
In this Lightning Lesson, Daniel will share a preview of Superlinked Inference Engine - an open source software for self-hosting Small Language Models in your own cloud. Cut 95%+ of your managed LLM API cost, gain access to a wide catalog of esoteric and fine-tunable SOTA models, and regain security and control by self-hosting.
You'll learn from
Daniel Svonava
CEO at Superlinked
Trey Grainger
Founder at Searchkernel, Author "AI-Powered Search".
Doug Turnbull
Principal Search Consultant
Mon Mar 2·6:00 PM UTC
Agentic Search: What's the big deal?
Agents rewire how search works. Historically search interprets a few short keywords. These represent a complex paragraph-length information need. We've built enormous complexity to understand, parse, and rank against that tiny hint of what the user wants. Why rely on hints? Agents can work against the full info need. They add reasoning + tool use + code generation, completely revamping search.
You'll learn from
Doug Turnbull
Search and Snuggie Enthusiast
Trey Grainger
Founder @ Searchkernel, Author "AI-Powered Search"
Tue Mar 3·5:00 PM UTC
Are we all doing "retrieval" wrong?... a candid discussion
Modern retrieval usually focuses on static algorithms (BM25, vector, and hybrid search) and models (embeddings, cross-encoders, etc.), decomposing search into isolated features or tools (for an agent). But is this the wrong approach? We’ll discuss why this often hits a wall and when more systematic approaches with continuous feedback loops (user signals, UX, evals, business KPIs, etc.) are needed.
You'll learn from
Andreas Wagner
Co-Founder & CTO @ searchHub.io
Trey Grainger
Founder @ Searchkernel, Author "AI-Powered Search"
Doug Turnbull
Search and Snuggie Enthusiast
Wed Mar 4·5:00 PM UTC
Modern AI Search Techniques You Should Know
Modern retrieval includes techniques like Hybrid Search (BM25, Dense Vector Search, etc), Late Interaction, Wormhole Vectors, Agentic Search, Semantic Knowledge Graphs, Learning to Rank, Signals Boosting, Sparse Encoders, Active Learning, Cross-encoders, Tensors, SPLADE, Muvera, ... the list goes on. Join Trey and Doug for a quick quick rundown of some key AI Search techniques worth understanding.
You'll learn from
Trey Grainger
Founder @ Searchkernel, Author "AI-Powered Search"
Doug Turnbull
Search and Snuggie Enthusiast
Fri Mar 6·3:30 PM UTC
AI Search Fundamentals: Vector Spaces, Matching, & Ranking
This session covers foundational concepts in information retrieval around matching and ranking. We'll cover sparse vs. dense vector spaces, lexical (BM25) vs. semantic (embedding) search, and matching and relevance ranking fundamentals. This session is free (open to all) and also provides the prerequisite background needed for anyone to join Trey & Doug's month-long "AI-Powered Search" course.
You'll learn from
Doug Turnbull
Principal ML Engineer in Search
Trey Grainger
Author, "AI-Powered Search", Founder @ Searchkernel