Ash Vardanian

StringZilla. Up to 100x faster strings for C, C++, CUDA, Python, Rust, Swift, JS, & Go, leveraging NEON, AVX2, AVX-512, SVE, GPGPU, & SWAR to accelerate search, hashing, sorting, edit distances, sketches, and memory ops 🦖

less_slow.cpp. Playing around "Less Slow" coding practices in C++ 20, C, CUDA, PTX, & Assembly, from numerics & SIMD to coroutines, ranges, exception handling, networking and user-space IO

NumKong. SIMD-accelerated distances, dot products, matrix ops, geospatial & geometric kernels for 16 numeric types — from 6-bit floats to 64-bit complex — across x86, Arm, RISC-V, and WASM, with bindings for Python, Rust, C, C++, Swift, JS, and Go 📐

ForkUnion. Lower-latency OpenMP-style minimalistic scoped thread-pool designed for 'Fork-Join' parallelism in Rust and C++, avoiding memory allocations, mutexes, CAS-primitives, and false-sharing on the hot path 🍴

SwiftSemanticSearch. Real-time on-device text-to-image and image-to-image Semantic Search with video stream camera capture using USearch & UForm AI Swift SDKs for Apple devices 🍏

StringWars. Comparing performance-oriented string-processing libraries for substring search, multi-pattern matching, hashing, edit-distances, sketching, and sorting across CPUs and GPUs in Rust 🦀 and Python 🐍

less_slow.py. Playing around "Less Slow" coding practices in Python, from numerical micro-kernels to coroutines, ranges, and polymorphic state machines

less_slow.rs. Playing around "Less Slow" coding practices in Rust, from numerical micro-kernels to coroutines, ranges, and polymorphic state machines

ParallelReductionsBenchmark. Thrust, CUB, TBB, AVX2, AVX-512, CUDA, OpenCL, OpenMP, Metal, and Rust - all it takes to sum a lot of numbers fast!

usearch-images. Semantic Search demo featuring UForm, USearch, UCall, and StreamLit, to visual and retrieve from image datasets, similar to "CLIP Retrieval"

SmashTable. If only std::set was a DBMS: collection of templated ACID in-memory exception-free thread-safe and concurrent containers in a header-only library

RetriEval. Benchmark suite that compares vector search engines against each other on billion-scale datasets, from in-memory HNSW libraries like USearch, FAISS, & cuVS to managed databases like Qdrant, Weaviate, LanceDB, & Redis – built for reproducibility

PyBindToGPUs. Parallel Computing starter project to build GPU & CPU kernels in CUDA & C++ and call them from Python without a single line of CMake using PyBind11

StringTape. Apache Arrow-compatible space-efficient "tape" class in pure Rust to be used with StringZilla for GPU, NUMA, and disk transfers of variable length strings

NetworkXternal. NetworkX-like Python experience for Postgres, SQLite, MongoDB, and Neo4J

TinySemVer. Tiny Semantic Versioning (SemVer) library with LLMs and GitHub CI, that doesn't depend on 300K lines of JavaScript code and fits in a single Python file

LibSee. Link to this library and it will log all the LibC functions you are calling and how much time you are spending in them!

JaccardIndex. Optimizing bit-level Jaccard Index and Population Counts for large-scale quantized Vector Search via Harley-Seal CSA and Lookup Tables

ScalingElections. GPU-accelerated Schulze voting method in Python, Numba, CUDA, and Mojo 🔥, using ideas from Algebraic Graph Theory

abusing-vector-search. Example of using Vector Search algorithms for non-traditional workloads, like GIS, stock prices, and sets

MongooseMiner. Documentation retrieval system to help LLMs navigate less-popular (yet often more powerful) Python libraries

SpaceV. Billion-scale Semantic Search dataset derived from Microsoft SpaceV for Vector Search benchmarks with smaller subsets

HashTableBenchmark. A simple cross-platform speed & memory-efficiency benchmark for the most common hash-table implementations in the C++ world

vector-dossier. Vector Dossier is a CLI tool that statically analyzes vectorization depth of programs and libraries

HashEvals. Minimalistic Rust toolkit for hash function quality analysis. Tests avalanche effect, differential patterns, and statistical distribution across variable-length n-grams.

extrapolaTED. Bringing TED experiences to every topic with Gen AI

TenPack. Fast Tensors Packaging library for text, image, video, and audio data compatible with PyTorch, TensorFlow, & NumPy 🖼️🎵🎥 ➡️ 🧠

minSH. A minimal Python re-implementation of the A* with seed heuristic for exact global alignmnet (edit distance) in near-linear time

ucall. Remote Procedure Calls - 50x lower latency and 70x higher bandwidth than FastAPI, implementing REST & JSON-RPC over io_uring and SIMDJSON ☎️

acid-redis. Tiny Redis-like Persistent ACID Store on RocksDB with JSON-RPC using UCall and UStore

HaversineSimSIMD. Staging area for Haversine distance computations in SimSIMD and USearch

PolyglotBot. Bot we've build for the Poe.com hackathon at the AGI house to gather results from multiple LLMs and trigger specialized models on-demand

CppNeuralSTL. Simple neural network models from scratch using only C++ STL

USearchBench.java. Apache Spark and Unum USearch integration example benchmarking distributed Vector Search against Lucene and OpenSearch

VectorHub. VectorHub is a free, open-source learning website for people (software developers to senior ML architects) interested in adding vector retrieval to their ML stack.

PyScrapeIg. Instagram scraper with proper pagination, that can collect posts, likes, comments and a lot more.

WolframSummer19. Quantifying the influence of urban design on quality of life.

xl-sum. This repository contains the code, data, and models of the paper titled "XL-Sum: Large-Scale Multilingual Abstractive Summarization for 44 Languages" published in Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021.

usearch. Vector Search engine in a single file - supporting C++ 11, C 99, Python 3, JavaScript, Rust, Java, GoLang, Objective-C, Swift, Wolfram, Docker, and WebAssembly 🔍

sentence-transformers. Multilingual Sentence & Image Embeddings with BERT

uform. Pocket-Sized Multimodal AI for content understanding and generation across multilingual texts, images, and 🔜 video, up to 5x faster than OpenAI CLIP and LLaVA 🖼️ & 🖋️

liburing. Library providing helpers for the Linux kernel io_uring support

lean4. Lean 4 programming language and theorem prover

awesome-vector-search. Collections of vector search related libraries, service and research papers

kdtree-rs. K-dimensional tree in Rust for fast geospatial indexing and lookup

llvm-project. The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.

open-data-registry. A registry of publicly available datasets on AWS

langchain. 🦜🔗 Build context-aware reasoning applications

usearch-langchain-js. Using USearch vector-search engine for RAG with LangChain

JsParseWiki. Wiki dump importer for MongoDB

probabilistic-collections-rs. Rust

nltk. NLTK Source

sse-popcount. SIMD (SSE) population count --- http://0x80.pl/articles/sse-popcount.html

conan-center-index. Recipes for the ConanCenter repository

oneTBB. oneAPI Threading Building Blocks (oneTBB)