Build your own vector database
Doug Turnbull
Led Search + ML at Reddit and Shopify
This course is popular
3 people enrolled last week.
Make embedding retrieval an asset, not a headache
In modern search, embeddings answer questions. They go beyond simple keyword matching to find semantically similar answers. Increasingly: RAG, recommendations, and traditional search need zero-in on what's relevant. In this class we'll live-code core data structures that live behind search systems like Pinecone, Weaviate, Turbopuffer, QDrant, Vespa, Elasticsearch etc etc etc.
By building your own vector database, you'll be better equiped to work with production vector search systems. You'll have first-hand experience with the knobs to turn to improve performance and develop a robust hybrid + vector search system
What you’ll learn
Save cost, improve your vector database performance, and be a savvier vector DB customer by learning core retrieval data structures
Industry best practices in evaluating a vector database for latency and recall
The pros/cons of graph vs clustering approaches to vector search
Learn how vector databases turn 1000s of floats into a compact representation.
The use cases that just call for naive, in-memory brute-force search. Or just a vector index library.
Workshop agenda
Benchmarking vector search
How we think through vector database stats: recall, latency, throughput
Building HNSW algorithm from scratch
Hands on developing the core graph-based algorithm behind search engines: HNSW
Filtering vector search
How to enhance vector algorithms to filter based on metadata
Optimizing with quantization
How to use standard quantization techniques to reduce memory, improve speed, without sacrificing recall
Layering in hybrid search
Lexical retrieval isn't obsolete - we'll talk about approaches to combining lexical and vector search into a single solution
Learn directly from Doug
Doug Turnbull
Search at Reddit, Shopify, Wikipedia
Who this workshop is for
Infrastructure teams - anyone tasked to squeeze the most performance out of a high-scale retrieval system
Search developers - anyone that needs to build relevant, fast vector search for RAG or agentic search applications
AI engineers - need to find that relevant context? Anyone who needs to find context to answer questions from AI
Prerequisites
Python programming
Comfort coding in Python is recommended. The course encourages you to code-along. Though you can still get value following the instructor.
You know what an embedding is
We'll describe how to search embeddings at scale. We assume you know what an embedding is
Numpy / Pydata optional, but useful
We represent vectors as numpy arrays of floats, as numpy provides several useful basic features for handling floats
What's included
Live sessions
Learn directly from Doug Turnbull in a real-time, interactive format.
Lifetime access
Go back to course content and recordings whenever you need to.
Community of peers
Stay accountable and share insights with like-minded professionals.
Certificate of completion
Share your new skills with your employer or on LinkedIn.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
Frequently asked questions
Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$600
USD
5 days left to enroll
1–5pm EDT