skills.rest
Orchestra-Research
Orchestra-ResearchOFFICIAL
Explore all skills by this author →
GitHub

qdrant-vector-search

Implements vector similarity search and hybrid retrieval for RAG systems using Qdrant.

13.0k|930|Updated Nov 3, 2025
Search Tags:#semantic search#embeddings#rag#vector database#vector search#hybrid search#qdrant
One-click install
npx skills add https://github.com/Orchestra-Research/AI-research-SKILLs --skill qdrant-vector-search-orchestra-research
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: qdrant-vector-search
Source: https://github.com/Orchestra-Research/AI-research-SKILLs/tree/main/15-rag/qdrant
Command: npx skills add https://github.com/Orchestra-Research/AI-research-SKILLs --skill qdrant-vector-search-orchestra-research

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires qdrant-client>=1.12.0, and includes references (resource) components.

What problem does it solve? Building production RAG and semantic search systems requires fast nearest neighbor search over embeddings with metadata filtering, which plain databases and in-memory indexes cannot deliver at scale. ## Core Features & Use Cases - Vector Search with Filtering: Perform HNSW-based nearest neighbor search combined with rich payload filters, including geo, full-text, and nested conditions. - Hybrid and Multi-Vector Search: Combine dense and sparse vectors with Reciprocal Rank Fusion, and store multiple named vectors per point. - Production Scaling: Use scalar, product, and binary quantization, sharding, replication, snapshots, and collection aliases for large-scale deployments. - Use Case: Index a knowledge base of documents embedded with sentence-transformers, then retrieve the top-k relevant chunks filtered by category to ground an LLM's answers in a RAG pipeline. ## Quick Start Use the qdrant skill to create a collection, upsert my document embeddings, and run a filtered similarity search against my local Qdrant instance.

Frequently Asked Questions about qdrant-vector-search

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

FAQPage Schema
How do I perform vector similarity search with filtering in Qdrant?▼

Use the qdrant-client search method with a query_vector and a query_filter containing must, must_not, or should conditions on payload fields. Create payload indexes on filtered fields like category or timestamp to keep filtered searches fast.

Qdrant vs FAISS vs Pinecone for RAG: which should I use?▼

Qdrant fits production RAG needing hybrid search, metadata filtering, and on-premise control. FAISS suits maximum raw speed in research or batch processing, Pinecone fits fully managed zero-ops preferences, and Chroma fits simpler embedded use cases.

Does Qdrant support hybrid search with dense and sparse vectors?▼

Yes, Qdrant supports named dense vectors alongside sparse vectors such as BM25 or SPLADE in the same collection. You can combine both with prefetch queries and Reciprocal Rank Fusion using FusionQuery.

Why is my Qdrant search slow when using filters?▼

Slow filtered search usually means missing payload indexes on the filtered fields. Create a keyword or integer index with create_payload_index, and consider enabling scalar quantization and gRPC for additional speed.

How do I reduce Qdrant memory usage for large collections?▼

Enable scalar, product, or binary quantization for 4x to 32x memory reduction, and set on_disk_payload and on_disk HNSW options to move data off RAM. Use rescoring during search to recover accuracy lost to quantization.

Can Qdrant run as a distributed cluster with replication?▼

Yes, Qdrant uses Raft consensus for distributed coordination with configurable shard_number and replication_factor per collection. You can tune write consistency with WriteOrdering and read consistency levels like majority or all.

Related Skills

Explore More Software Engineering Skills→
MedTiLabMedTiLab
5

sentence-transformers

Generate sentence, text, and image embeddings for semantic search and RAG.

Community
Intermediate
JCETools-PetraJCETools-Petra
16

ai-llm-engineering

Design RAG pipelines with chunking, embedding, retrieval, and evaluation metrics.

Community
Advanced
agentsopeagentsope
287

agentsop-reranker-stage

Rerank bi-encoder candidate sets with cross-encoders to improve RAG precision.

Community
Advanced
jillescajillesca
3

langchain-rag

Build end-to-end RAG pipelines with LangChain document loaders, embeddings, and vector stores.

Community
Advanced