skills.rest
ricable
ricableCOMMUNITY
Explore all skills by this author →
GitHub

faiss

Build fast nearest-neighbor search over dense vectors with FAISS indexes.

Updated Feb 15, 2026
Search Tags:#vector search#similarity search#embedding retrieval#faiss#gpu acceleration#nearest neighbors
One-click install
npx skills add https://github.com/ricable/mcai --skill faiss-ricable
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: faiss
Source: https://github.com/ricable/mcai/tree/main/.agents/skills/faiss
Command: npx skills add https://github.com/ricable/mcai --skill faiss-ricable

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

FAISS solves fast similarity search over large dense vector collections when exact nearest-neighbor lookup must scale to millions or billions of embeddings. It helps teams retrieve semantically similar items, rank recommendations, and cluster embeddings without building a full database layer.

Core Features & Use Cases

  • Exact and approximate nearest-neighbor search with Flat, IVF, HNSW, and PQ indexes.
  • GPU acceleration and multi-GPU scaling for high-throughput embedding workloads.
  • Save and reload trained indexes for offline retrieval, recommendation, and RAG pipelines.
  • Use case: build a semantic search service that indexes document embeddings and returns the most similar passages to a query in real time.

Quick Start

Use this Skill to choose the right FAISS index for your dataset, create the vector store, and search for nearest neighbors efficiently.

Frequently Asked Questions about faiss

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

FAQPage Schema
What is the best way to build fast similarity search over millions of dense vectors?▼

Fast similarity search over millions of dense vectors is built by selecting and training a FAISS index, such as Flat, IVF, HNSW, or PQ, to retrieve exact or approximate nearest neighbors in milliseconds.

How do I choose the right FAISS index type for large-scale semantic retrieval?▼

Choosing the right FAISS index depends on your scale and latency requirements: Flat indexes provide exact nearest neighbor retrieval, while IVF, HNSW, and PQ indexes offer approximate similarity search optimized for millions to billions of embeddings.

Can I use GPU acceleration for embedding retrieval in RAG workflows?▼

Yes, GPU acceleration and multi-GPU scaling are supported for embedding retrieval in RAG workflows, enabling high-throughput nearest neighbor search and query tuning without building a full database layer.

How do I save and reload a trained vector index for offline recommendation pipelines?▼

You can save and reload trained FAISS indexes to support offline retrieval, clustering, and recommendation pipelines, ensuring persistent nearest neighbor search across different execution sessions.

Does vector search with FAISS require a separate database layer for billion-scale embeddings?▼

No, FAISS handles nearest-neighbor search over billions of dense vectors directly without requiring a full database layer, allowing teams to retrieve semantically similar items and rank recommendations efficiently.

Related Skills

Explore More Software Engineering Skills→
an8079an8079
4

embedding-generation

Train and optimize embedding models for text, images, and multimodal data.

Community
Advanced
ldzhhxxldzhhxx
1

qdrant-vector-search

Index vectors from data/vectors.csv and search for similar vectors.

Community
Advanced
vecearvecear

rag-pipeline-builder

Design retrieval-augmented generation pipelines with chunking, embeddings, and reranking.

Community
Advanced
zeronx798zeronx798

pinecone

Create serverless indexes and run hybrid vector searches with metadata filtering.

Community
Intermediate