skills.rest
leonardoteodoroo
leonardoteodorooCOMMUNITY
Explore all skills by this author →
GitHub

vector-index-tuning

Tune HNSW parameters and quantization strategies to optimize vector search latency, recall, and memory.

Updated Feb 3, 2026
Search Tags:#recall#latency#quantization#hnsw#vector-search#index-tuning
One-click install
npx skills add https://github.com/leonardoteodoroo/amino-advanced --skill vector-index-tuning-leonardoteodoroo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: vector-index-tuning
Source: https://github.com/leonardoteodoroo/amino-advanced/tree/main/.agent/skills/vector-index-tuning
Command: npx skills add https://github.com/leonardoteodoroo/amino-advanced --skill vector-index-tuning-leonardoteodoroo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Optimize vector index performance by tuning HNSW parameters, selecting quantization strategies, and scaling vector search infrastructure to meet latency, recall, and memory goals.

Core Features & Use Cases

  • HNSW parameter tuning (M, efConstruction, efSearch) to balance recall and latency.
  • Quantization strategy selection (FP32, FP16, INT8, PQ, binary) to reduce memory without sacrificing too much accuracy.
  • Memory and performance estimation across dataset sizes to guide deployment decisions.
  • Use Case: You operate a vector search service with hundreds of millions of vectors and need faster queries with controlled memory.

Quick Start

Tune HNSW and quantization settings for your current dataset to achieve the target latency and recall.

Frequently Asked Questions about vector-index-tuning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I tune HNSW parameters to improve vector search recall and reduce latency?▼

Tune HNSW parameters like M, efConstruction, and efSearch to balance vector search recall and latency. Increasing efSearch generally improves recall at the cost of higher query latency, while adjusting M and efConstruction impacts index build time and memory consumption.

What is the best quantization strategy to reduce vector index memory usage?▼

Choosing the right quantization strategy, such as FP16, INT8, Product Quantization (PQ), or binary, reduces vector index memory usage. FP32 offers full precision, while lower-bit formats trade slight accuracy drops for significant memory savings across large datasets.

Can I estimate memory impact before deploying a vector search index with billions of vectors?▼

Yes, you can estimate the memory impact for different HNSW and quantization configurations across dataset sizes ranging from thousands to billions of vectors. This helps guide deployment decisions to ensure controlled memory usage and target latency.

Why does my vector search latency spike when I increase efSearch for higher recall?▼

Vector search latency spikes when increasing efSearch because the algorithm explores more nodes in the HNSW graph to find accurate matches. This deeper graph traversal improves recall but directly increases query processing time per request.

Does vector index tuning work for production deployments with hundreds of millions of vectors?▼

Vector index tuning is applicable to production vector search deployments with hundreds of millions of vectors. By configuring HNSW parameters and selecting appropriate quantization strategies, you can achieve faster queries with controlled memory usage.

Related Skills

Explore More Data & Analytics Skills→
emilneuraz-aiemilneuraz-ai

vector-index-tuning

Tune HNSW parameters and quantization to optimize vector index latency and recall.

Community
Intermediate
ccfccf

vector-index-tuning

Tune HNSW parameters and quantization for large-scale vector search indexes.

Community
Advanced
gemini-cli-extensionsgemini-cli-extensions
42

cloud-sql-postgres-vectorassist

Automate vector search configuration and optimization on Cloud SQL for PostgreSQL.

Official
Intermediate
ruiyangruiyiruiyangruiyi

faiss

Builds and queries optimized FAISS indexes for fast k-NN similarity search on dense vectors.

Community
Intermediate