advanced~2h
Redis — Vector Search & Semantic Caching for LLM Applications
Redis's in-memory speed turns out to matter for a newer workload too: storing and searching embeddings. This chapter covers RediSearch's vector similarity capabilities and the specific pattern of using Redis as a semantic cache in front of an LLM — caching based on what a prompt means rather than its exact text.
Learning objectives
- Explain how RediSearch's vector field type and index enable similarity search inside Redis.
- Store an embedding alongside a document in a Redis hash and query it for nearest neighbors.
- Explain the difference between exact-match caching and semantic caching for LLM prompts.
- Implement a basic semantic cache lookup using a similarity threshold rather than an exact key match.
- Reason about the cost/latency trade-offs semantic caching is actually solving for in an LLM application.
This is a Pro chapter
Sign in, then upgrade to Pro or Power to unlock this and the full Databases Mastery library.
Redis — Vector Search & Semantic Caching for LLM Applications