Using Vector Databases: Practical Advice for Production - LLMs in Prod Conference

Discover practical advice for using vector databases in production environments during this 30-minute conference talk from the LLMs in Prod Conference. Explore various use cases for vector databases with large language models, including information retrieval, conversational memory for chatbots, and semantic caching. Delve into the less flashy but crucial aspects of implementing these technologies, such as prompt engineering, text chunking, compliance considerations, on-premise solutions, embedding model changes, index types, A/B testing, cloud platform selection, deployment strategies, feature injection, and tool comparisons. Gain insights from a year's worth of Redis deployments for AI use cases, condensed into a comprehensive overview. Learn about similarity searches, design patterns, context retrieval, feature injection, query optimization, guard rails, long-term memory, common mistakes, and index management. Benefit from the expertise of Sam Partee, a Principal Engineer guiding AI efforts at Redis, as he shares valuable knowledge on integrating vector databases into ML pipelines for feature storage, search, and inference workloads.

Intro
Similarity searches
Redis
Design Patterns
Context Retrieval
Hide
Feature Injection
Query Get
Guard Rails
Longterm Memory
Common Mistakes
Laziness
Index Management
Metadata