Menü
Reduce retrieval latency with caching, batching, and index-level optimization.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
Use this skill when building, debugging, or improving Retrieval-Augmented Generation systems, including chunking, vector database selection, hybrid search, reranking, multimodal RAG, code documentation RAG, retrieval latency, and production RAG architecture.
Chunk nested documents into parent-child levels so retrieval can move from broad sections to fine-grained passages.
Use semantic boundaries and embedding similarity to chunk text for higher-relevance retrieval.
Route RAG chunking decisions across semantic, hierarchical, sliding-window, contextual-header, and framework-selection strategies.
Use overlapping windows to preserve context across chunk boundaries while controlling retrieval size.
Route RAG performance work for latency, caching, indexing, filtering, batching, and query optimization.
| name | optimize-retrieval-latency |
| title | Optimize Retrieval Latency |
| description | Reduce retrieval latency with caching, batching, and index-level optimization. |
| allowed-tools | ["Read","Grep","Glob","Bash"] |
| category | performance-optimization |
| tags | ["latency","performance","caching","indexing","optimization"] |
Optimizing RAG retrieval latency is critical for production applications where user experience depends on fast response times. This skill covers techniques to reduce query latency at every stage: indexing, embedding generation, search execution, and result processing.
RAG systems often have latency issues:
Establish baseline measurements before optimization.
Why: You can't optimize what you don't measure. Profiling reveals the actual bottlenecks rather than assumptions.
Reduce the overhead of generating query embeddings.
Why: Embedding generation is often the bottleneck - caching reduces this overhead for repeated queries.
Tune indexing and search parameters for latency.
Why: Index parameters directly impact search latency, lower ef values reduce search time at the cost of recall.
Cache at multiple levels to reduce redundant work.
Why: Multi-level caching reduces computation for repeated queries without sacrificing freshness for unique queries.
Stream results and minimize post-processing overhead.
Why: Streaming reduces memory usage and allows users to see results sooner, improving perceived latency.
Prioritize and batch queries for better resource utilization.
Why: Query prioritization ensures important queries get resources first and reduces contention.
Useful references include the Qdrant production guide, Qdrant snapshots concept, ANN Benchmarks, and AWS caching best practices.
