Top Builders

Explore the top contributors showcasing the highest number of app submissions within our community.

Qdrant

Qdrant is a high-performance, open-source vector database and search engine designed specifically for vector similarity search. Written in Rust, Qdrant offers fast and reliable performance even under high load, making it an ideal choice for applications that demand speed and scalability. With Qdrant, you can turn your embeddings or neural network encoders into powerful, full-fledged applications for a wide range of use cases including matching, search, recommendation, and complex operations on large datasets.

Qdrant's comprehensive API and extended filtering support make it well-suited for various applications, including faceted search, semantic-based matching, and hybrid search combining vector similarity with keyword matching. The platform offers both self-hosted and managed cloud solutions, providing flexibility for different deployment scenarios.

General
Release DateOctober 29, 2021
AuthorQdrant
TypeVector Database & Search Engine
LanguageRust
LicenseApache License 2.0
Latest Version1.15.x (July 2025)
GitHub Stars19,000+ (as of 2025)

Core Features

High-Performance Vector Search

  • HNSW (Hierarchical Navigable Small World) indexing algorithm
  • Sub-millisecond search latency for millions of vectors
  • Handles billions of vectors with horizontal scaling
  • SIMD hardware acceleration for optimal CPU utilization

Advanced Quantization Techniques

  • Scalar Quantization: Reduces memory usage by 4x with minimal accuracy loss
  • Binary Quantization: Up to 32x compression with 40x speed improvement
  • Product Quantization: Higher compression ratios for high-dimensional vectors
  • 1.5-bit and 2-bit Quantization: New compression options balancing efficiency and accuracy
  • Asymmetric Quantization: Different encoding for stored vectors vs. queries

Hybrid Search Capabilities

  • Dense Vectors: Traditional semantic similarity search using neural embeddings
  • Sparse Vectors: Keyword-based search with neural network weighting (BM25, SPLADE)
  • Combined Search: Unified queries across both dense and sparse vectors in single collections
  • BM42 Algorithm: Proprietary hybrid search combining vector and keyword methods

Comprehensive Filtering & Payload Support

  • Rich JSON payload attachment to vectors
  • Advanced filtering conditions (must, should, must_not clauses)
  • Geo-location, numerical ranges, and full-text filtering
  • Custom business logic implementation on top of similarity matching

Latest Enhancements (2024-2025)

Qdrant 1.15.x (July 2025)

  • Smarter Quantization: Enhanced 1.5-bit and 2-bit quantization options
  • Advanced Full-Text Filtering: Improved text search capabilities
  • Performance Optimizations: Better resource utilization and faster processing

Qdrant 1.14.x (April 2025)

  • Score-Boosting Reranker: Blend vector similarity with custom rules and context
  • Incremental HNSW Indexing: Build indexes gradually as data arrives
  • Optimized Batch Search: Parallel processing for large query batches
  • Memory Optimization: Reduced RAM usage for large datasets

Qdrant Cloud Inference (July 2025)

  • Unified Embedding Generation: Generate, store, and index embeddings in single API calls
  • Multimodal Support: Text and image embedding models integrated natively
  • Supported Models: MiniLM, SPLADE, BM25, Mixedbread Embed-Large, CLIP
  • Free Tier: 5 million tokens per model monthly, unlimited BM25 tokens

Deployment Options

Self-Hosted (Open Source)

  • Full control over infrastructure and data
  • Horizontal scaling through sharding and replication
  • Zero-downtime rolling updates and dynamic scaling
  • Available on GitHub with comprehensive documentation

Qdrant Cloud (Managed Service)

  • Fully managed SaaS solution across AWS, Google Cloud, and Azure
  • Automatic scaling, monitoring, and maintenance
  • Built-in backups and disaster recovery
  • Simple setup with production-ready clusters in minutes

Hybrid Cloud

  • Deploy across diverse environments for performance and compliance
  • Integrate on-premise and cloud infrastructure
  • Maintain data residency requirements while leveraging cloud benefits

Private Cloud

  • Full Kubernetes-based deployment on any infrastructure
  • Complete control and customization for enterprise requirements
  • Air-gapped deployments for maximum security

Key Use Cases

Retrieval-Augmented Generation (RAG)

  • Semantic search over knowledge bases and documents
  • Context retrieval for large language models
  • Enterprise knowledge management and Q&A systems

Recommendation Systems

  • Product recommendations based on user behavior
  • Content discovery and personalization
  • Similar item matching across catalogs

Multimodal Search

  • Combined text and image search capabilities
  • Video content indexing and retrieval
  • Cross-modal similarity matching

Enterprise Search

  • Semantic document search with filtering
  • Compliance and regulatory document retrieval
  • Customer support knowledge bases

Anomaly Detection

  • Outlier detection in high-dimensional data
  • Fraud detection and security monitoring
  • Quality control and process optimization

Technical Specifications

Performance Characteristics

  • Throughput: Up to 10,000+ queries per second
  • Latency: Sub-millisecond search response times
  • Scale: Billions of vectors with horizontal scaling
  • Memory Efficiency: Up to 97% RAM reduction with quantization

Supported Distances

  • Cosine similarity
  • Euclidean (L2) distance
  • Dot product
  • Manhattan (L1) distance

Integration Ecosystem

  • Embeddings: OpenAI, Cohere, Hugging Face, Sentence Transformers
  • Frameworks: LangChain, LlamaIndex, Haystack
  • Languages: Python, Rust, JavaScript, Go, .NET
  • Platforms: Docker, Kubernetes, cloud providers

Recent Achievements

  • 250+ million installs across all open-source packages
  • Featured in The Forrester Waveâ„¢: Vector Databases, Q3 2024
  • Named one of Europe's top 10 startups in Sifted's 2025 B2B SaaS Rising 100
  • Powers AI applications at Tripadvisor, HubSpot, and Deutsche Telekom
  • Active community of 6,000+ members on Discord

Tutorials

Great tutorials on how to build with Qdrant

Resources

Getting Started

Community & Support

Qdrant represents the cutting edge of vector database technology, combining high performance with developer-friendly APIs and comprehensive feature sets. Whether building semantic search, recommendation systems, or complex AI applications, Qdrant provides the infrastructure needed to scale from prototype to production with confidence.

Edit on GitHub

Qdrant AI technology page Hackathon projects

Discover innovative solutions crafted with Qdrant AI technology page, developed by our community members during our engaging hackathons.

RocmPilot Studio — CUDA→AMD Migration Cockpit

RocmPilot Studio — CUDA→AMD Migration Cockpit

AMD's MI300X wins on price/performance, but what keeps enterprises on CUDA isn't silicon — it's migration uncertainty. A typical AI codebase is CUDA-first top to bottom: hardcoded cuda device logic, nvidia/cuda base images, CUDA-pinned wheels, and the part everyone fears — custom kernels with warp-level assumptions. Tools like hipify translate the easy majority, but nobody can tell you up front how ready a whole codebase is, or where the hard work hides. RocmPilot Studio is the command center for that migration. Paste a repo URL and it runs a six-stage pipeline: SCAN → PLAN → PATCH → VALIDATE → REPORT. • Scan: a deterministic engine (no LLM — findings can't be hallucinated) detects blockers with file, line and severity, plus a kernel-risk classifier that names the hard 20% in AMD vocabulary: 32-lane warp shuffles vs 64-lane wavefronts, warpSize hazards, WMMA→rocWMMA, CUTLASS→Composable Kernel, cuBLAS→hipBLAS, NCCL→RCCL — at repository scale. • Plan: a multi-model agent orchestra (all AMD-hosted via Fireworks AI) — DeepSeek drafts the prioritized plan, then GLM, a different model, independently critiques it before you see it. Live-streamed agent trace in the UI. • Patch: safe auto-fixes as patch.diff, plus a ready-to-run ROCm Dockerfile, smoke test and benchmark — every change explained from the real diff. Download the fully patched repo as a zip. • Validate: smoke test + benchmark executed on real AMD hardware (passed on Radeon gfx1100, ROCm 7.2, PyTorch 2.9.1, ~0.84 ms inference; replays are always labeled). On failure, a research agent (Kimi) returns a cited, RAG-grounded fix from a ROCm/HIP knowledge base in Qdrant. • Report: an honest, blocker-weighted readiness score (sample repo: 37 → 72 → 86) and an exportable report. Fallback-safe at every layer — no API key, no vector DB, no GPU still yields a working, honest run. 160+ deterministic tests. Built by a two-person team with a protected-branch PR workflow.

RasoSynthTune: Text-to-Dataset & Fine-Tuning

RasoSynthTune: Text-to-Dataset & Fine-Tuning

ML teams fine-tuning models spend immense time hand-curating data or blindly trusting API-generated data that silently poisons training runs with hallucinations. RasoSynthTune solves this bottleneck as an autonomous multi-agent data refinery that converts raw, unstructured documents into verified fine-tuning datasets. The platform ensures quality through a rigorous three-layer verification shield: consensus-based LLM judge scoring, iterative self-refinement where low-scoring samples are dynamically revised using the judge's own critiques, and a Human-in-the-Loop (HITL) manual review API. Furthermore, the system actively truncates hallucinated claims in generated multi-turn dialogues. Crucially, RasoSynthTune provides genuine hardware utilization rather than just standard API routing. Heavy multi-agent workflows and dynamic inference routing natively prioritize Fireworks AI, executing Google Gemma models (gemma2-9b-it and gemma2-27b-it) directly on AMD Instinct GPUs. The platform takes this further with an on-metal LoRA distillation loop, utilizing ROCm to fine-tune a smaller model directly on AMD hardware using the pipeline's own rigorously verified output data. Built for enterprise scale, the entire architecture—featuring a LangGraph multi-agent engine, FastAPI router, PostgreSQL, Redis, Qdrant vector storage, and a React/Next.js visual workflow studio—is fully containerized and features OpenTelemetry and Prometheus observability. It is engineered to deploy instantly on the AMD Developer Cloud, delivering transparent datasets (JSONL, Parquet) complete with hardware and model provenance cards.

CyLnk Learn

CyLnk Learn

By 2026, the world faces a 4.8M cybersecurity workforce gap. Traditional training fails: passive videos and certification mills produce paper engineers, not defenders. Meanwhile, 95% of cyberattacks target SMEs that can't afford $200K/year experts. Introducing CyLnk Learn: an AI-powered adaptive platform that fixes this by combining three pillars: 1. Probabilistic Mastery: Bayesian Knowledge Tracing (BKT) + SM-2 models every skill as a probability. Every next step is chosen by math, not a fixed table. 2. Vision-Aware AI Tutor: A LangGraph state machine over a RAG corpus. Paste a screenshot of a failing terminal, and our multimodal AI instantly diagnoses the exploit. 3. Live Sandboxed Labs: On-demand Docker/Podman exploit environments, isolated per learner, auto-graded, and evicted after 30 mins idle. Powered by AMD & Fireworks AI: We use a hybrid inference architecture running natively on AMD Instinct GPUs. We run lightweight Gemma models locally via AMD ROCm for zero data leakage, and route complex tasks through the Fireworks AI API. We implement smart model routing: Gemma 4B for intent, 12B for pedagogy, and 31B for vision analysis. This "right tool for the job" approach maximizes intelligence while minimizing inference costs. Market & Mission: Our beachhead is B2C learners in developing nations (Bangladesh, Nigeria, Vietnam) and under-defended SMEs globally. By making high-quality, personalized cybersecurity education affordable, we turn motivated learners into job-ready defenders, creating a remote talent pipeline for SMEs that desperately need them. CyLnk Learn isn't just a hackathon project; it's a battle-tested, validated startup (Winner of Infinity AI BuildFest 2026, Runner-Up Technopreneurship 2026). Learn by exploiting. Not by watching.

ISEEKAIGO

ISEEKAIGO

Most AI companions today are built around static prompts layered on top of large language models. They can generate impressive responses, but they rarely evolve beyond the conversation in front of them. ISEEKAIGO is building the next generation of AI companionship by combining generative AI with persistent memory, human-inspired needs, autonomous behavior, and evolving cognition. Instead of simply responding, our AI learns from experience, self-learning, adapts over time, and develops its own behavioral context, making every interaction more personal and meaningful. Our vision extends beyond creating a better chatbot. We are building a foundation for Human-like AI that can interact with people over long periods while becoming increasingly capable through experience. By breaking this vision into scalable milestones, we aim to transform AI companionship into a platform for future autonomous AI applications such as building physical-AI core. We believe this project represents a meaningful step toward the future of Human-like AI, combining ambitious research with the ability to ship real products and bring advanced AI experiences to millions of users.

Cortex: Industrial AI Knowledge Platform

Cortex: Industrial AI Knowledge Platform

Problem: Industrial organizations maintain thousands of pages of manuals, SOPs, maintenance guides, equipment specifications, engineering reports, and compliance documents. Critical knowledge is fragmented across PDFs, making retrieval slow and preventing engineers from answering questions that require combining information from multiple documents. Traditional RAG systems retrieve isolated chunks of text but fail to capture relationships between assets, procedures, failures, regulations, and components. Solution: Cortex is an Industrial Knowledge Intelligence Platform that transforms unstructured documents into a living knowledge graph and enables engineers to interact with it using a multi-agent AI system. Instead of retrieving only text, Cortex combines: Layout-aware document understanding Knowledge graph construction Dense vector retrieval Graph traversal Multi-agent reasoning to answer complex industrial questions with complete citations and provenance. Key Features Layout-aware PDF parsing Knowledge Graph generation Hybrid Retrieval (Dense + Graph) Multi-Agent AI Copilot Streaming responses Interactive graph exploration Explainable answers with citations Production-grade asynchronous ingestion JWT-secured backend Self-healing background workers AMD Compute Usage All machine learning workloads execute on AMD GPUs through AMD AI Notebooks. The unified inference gateway serves: ROCm + vLLM for LLM inference IBM Docling for PDF parsing and OCR FastEmbed for embedding generation The backend communicates with the gateway through OpenAI-compatible APIs without requiring model-specific code. Real-World Applications Manufacturing Energy Oil & Gas Chemical Plants Aerospace Industrial Compliance Asset Maintenance Technical Documentation Search

MemoryMint — Pay-per-Query AI Second Brain

MemoryMint — Pay-per-Query AI Second Brain

MemoryMint is a decentralised, pay-per-query AI second brain built for AMD Developer Hackathon ACT II. PROBLEM Knowledge workers generate dozens of documents weekly but can never search across them. Existing tools require subscriptions, centralised APIs, and don't support AI agent access natively. SOLUTION Upload any PDF or TXT file → MemoryMint chunks and embeds it using BAAI/bge-small-en-v1.5 on AMD Instinct MI300X with ROCm 7.2.4 → query in natural language → get cited answers. Every query is gated by X402 micropayments — $0.001 USDC on Base Sepolia, verified on-chain via x402.org/facilitator. KEY FEATURES - Own brain: upload files, chat with your knowledge, pay $0.001 per answer - Public brain sharing: publish your brain at /brain/{id}, earn USDC every time someone queries it - Agent-to-agent payments: AI agents discover price via /payment-info, pay autonomously via EIP-3009 and CDP SDK — zero human interaction - Earnings dashboard: real-time USDC earned, per-source breakdown, transaction feed with Basescan links - Python SDK: pip install memorymint → 5 lines to query any public brain AMD INTEGRATION - BAAI/bge-small-en-v1.5 runs on AMD Instinct MI300X via ROCm 7.2.4 - 6.5x embedding throughput vs CPU — 2914.1 chunks/sec on MI300X vs 448.2 chunks/sec on CPU (see benchmark.py) - GPU auto-detected at startup — falls back to CPU gracefully for local dev X402 INTEGRATION - HTTP 402 middleware gates /query and /brain/{id}/query endpoints - Browser wallets (MetaMask, Coinbase Wallet) pay via ERC-20 USDC transfer - AI agents pay via EIP-3009 off-chain signatures using Coinbase CDP SDK - x402.org/facilitator verifies all payments before requests are processed TECH STACK FastAPI · Qdrant · BAAI/bge-small-en-v1.5 · Groq llama-3.3-70b · Next.js App Router · wagmi v3 · viem · X402 · Base Sepolia · AMD Instinct MI300X · ROCm 7.2.4