Architectural Decision Guide: When to Use Vector Search (and When NOT To)
System design criteria for selecting vector search versus lexical, exact-match, or relational indexing architectures.
8 docs tagged with "system-design"
View all tagsDocument Chunking Strategies for Production RAG Pipelines
Architectural guide to text chunking strategies: fixed-size, recursive character, semantic chunking, and hierarchical parent-document retrieval.
Redis Agent Memory Basics: Stateful LLM Architectures
Architecting stateful LLM agents with short-term session continuity and durable long-term memory using Redis and LangGraph.
Retrieval-Augmented Generation (RAG): End-to-End System Architecture
Comprehensive engineering deep-dive into production Retrieval-Augmented Generation architectures, hallucination mitigation, and ingestion pipelines.
Vector Database Benchmarking: Throughput, Latency, and Recall Under Load
Rigorous benchmarking methodology for vector engines: analyzing QPS, P95/P99 latency, Recall@K curves, and concurrency stress testing.
Vector Database Internals: Storage Engines, Indexing, and Metadata Payloads
System internals of vector databases: indexing high-dimensional embeddings, coordinating metadata payloads, and managing CRUD lifecycles.
Vector Database Search: HNSW Algorithm Architecture & Deep-Dive
Comprehensive mathematical and algorithmic deep dive into Hierarchical Navigable Small World (HNSW) graphs for sub-millisecond vector retrieval.
What is a semantic cache？
System design of semantic caching architectures for LLMs: similarity thresholds, TTL policies, and cost reduction.