Explore the world of vector embedding-powered applications in production with this comprehensive conference talk from Conf42 Machine Learning 2023. Delve into the limitations of natural language processing and discover how vector embeddings offer superior solutions for search and recommendation systems. Learn about approximate nearest neighbors, content and user vector creation, and the implementation of query managers. Gain insights into building an MVP for vector-powered apps and understand the potential of a #vectorops platform. Investigate the role of generative AI, chat systems, and agent-based approaches in this evolving field. Connect with industry experts and expand your knowledge of cutting-edge machine learning techniques for practical applications.
Vector Ops: Running Vector Embedding-Powered Apps in Production
Overview
Syllabus
intro
preface
building vector-powered apps
what did we lose with language?
natural language is a bottleneck
natural language is ambiguous
vectors are better! mostly
search before vectors
recommendations before vectors
search & recommendations with vectors
approximate nearest neighbours?
it can be measured
building the content vectors...
...and user vectors in the same space!
let ann do the heavy lifting
query manager on top
what will you need to get started & MVP?
towards a #vectorops platform
what about generative ai?!
chat and chains!
what the frig?
agents & memory!
let's connect!
Taught by
Conf42
