NumPy

The fundamental N-dimensional array library powering Python's scientific computing stack.

Library
PyPI
v2.5.4
33,376 stars
BSD 3-Clause License

Repository Health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum. How we score it →
96 /100 Excellent
Development Activity 100
Maintenance 96
Community 88
Maturity 60
Momentum 40

Technical Analysis

AI-assessed by reading the actual repository — architecture, code quality, innovation, and documentation. How we score it →
90 /100 Excellent
Architecture 92
Code Quality 95
Innovation 88
Learning Curve 85

NumPy is the foundational library for numerical and scientific computing in Python, providing a fast, memory-efficient N-dimensional array object (ndarray) alongside broadcasting, vectorized universal functions (ufuncs), and tools for linear algebra, Fourier transforms, and random number generation. It underpins the vast majority of the Python data science and machine learning ecosystem — pandas, SciPy, scikit-learn, PyTorch, and TensorFlow all build directly on NumPy’s array protocol and C API.

Originally created in 2005 by merging the Numeric and Numarray projects, NumPy is now maintained by NumFOCUS and a large, active contributor base. Its core is implemented in C for performance, built with Meson and Cython bindings, while the Python-facing API exposes a broad, fully-typed surface (bundled .pyi stubs) for array manipulation, indexing, and interoperability with C/C++ and Fortran code via f2py.

What You Get

  • A fast, memory-efficient N-dimensional array object (ndarray) with broadcasting and vectorized operations
  • Comprehensive linear algebra, Fourier transform, and random number generation modules
  • C, C++, and Fortran interoperability via f2py and a stable C API for building extensions
  • Bundled type stubs (.pyi) for static typing support across the entire public API

Common Use Cases

  • Numerical simulations and scientific computing pipelines
  • Backing array/tensor operations for machine learning frameworks like PyTorch and TensorFlow
  • Data preprocessing and vectorized transformations ahead of pandas or SciPy analysis
  • Image and signal processing via array slicing, FFTs, and linear algebra routines

Under The Hood

Architecture - NumPy’s core (numpy/_core) is a C-implemented ndarray type and ufunc dispatch machinery; the Python layer wraps and extends it through dedicated submodules — numpy/lib, numpy/linalg, numpy/fft, numpy/random, numpy/ma, numpy/polynomial, numpy/matrixlib — each exposing a typed Python API backed by native code, with numpy/f2py bridging Fortran extensions into the same array protocol.

Tech Stack - The codebase is primarily Python (61%) and C (33%), with C++, Cython, and Fortran mixed in for performance-critical paths; the build system is Meson via meson-python (>=0.20.0) with Cython (>=3.1.0), targets Python >=3.12, and vendors several third-party numeric libraries (libdivide, Google Highway, x86-simd-sort, pocketfft) each tracked with its own license in pyproject.toml.

Code Quality - Every subpackage carries a colocated tests/ directory (numpy/_core/tests, numpy/linalg/tests, numpy/random/tests, etc.) run with pytest and hypothesis; ruff.toml and .clang-format enforce consistent Python and C style, pytest.ini configures test execution, and CI runs across both GitHub Actions and CircleCI for cross-platform and cross-architecture coverage.

API Design - The ndarray-centric API stays internally consistent across submodules — dtype and broadcasting semantics behave uniformly whether you’re in linalg, fft, random, or ma — and basic usage requires minimal boilerplate (import numpy as np; np.array(...)), though the project’s sheer API surface area and native build toolchain raise the bar for anyone compiling NumPy from source or contributing to its C internals.

Used by 81 apps in this directory

C
73%
AGPL 3.0

TDengine

Databases

25,187

A high-performance, open-source time-series database built in C for IoT, connected vehicles, and industrial monitoring workloads, with built-in stream processing, caching, and data subscription.

View details
98
Repo Health
71
Technical
65
Dependency
Built with
C 73%
C++ 20%
Updated 1 weeks ago
Go
95%
Apache 2.0

TiDB

AI Development · Databases

40,715

AI-Native Distributed SQL Database for Agentic Workloads

View details
97
Repo Health
66
Technical
68
Dependency
Built with
Go 95%
Updated 1 weeks ago
Python
73%
Apache 2.0

timesfm

AI Development · Analytics

35,277

A pretrained decoder-only foundation model by Google Research that delivers zero-shot time series forecasting with calibrated quantile prediction intervals — no training required.

View details
75
Repo Health
81
Technical
71
Dependency
Built with
Python 73%
HTML 14%
Jupyter Notebook 12%
Updated 1 months ago
Rust
71%
MIT

Trieve

AI Development · Developer Tools · Search

2,723

All-in-one self-hostable platform for hybrid search, RAG, recommendations, and analytics built on Rust and Qdrant.

View details
42
Repo Health
74
Technical
65
Dependency
Built with
Rust 71%
Updated 9 months ago
Python
72%
Apache 2.0

Unsloth

AI Assistants · AI Development

79,133

Run and fine-tune LLMs, diffusion, audio and embedding models on your own hardware, from a native desktop app, a browser UI, or a Python library.

View details
89
Repo Health
83
Technical
70
Dependency
Built with
Python 72%
TypeScript 24%
Updated 1 weeks ago
Python
96%
AGPL 3.0

VideoHighlighter

Video Editors

377

Local AI desktop tool that scores, explains, and cuts the moments that matter in raw footage — scene, motion, audio, object, and action detection, fully offline via Ollama and Whisper.

View details
83
Repo Health
84
Technical
71
Dependency
Built with
Python 96%
Updated 2 weeks ago
Python
46%
MIT

Voicebox

AI Development · Productivity · Voice AI

59,686

Clone voices, dictate anywhere, and give AI agents your voice — all locally.

View details
83
Repo Health
76
Technical
68
Dependency
Built with
Python 46%
TypeScript 42%
Updated 1 months ago
Python
48%
AGPL 3.0

VoiceStudio

Mcp · Music Audio · Voice AI

102,997

Open-source, fully local ElevenLabs alternative for voice cloning, voice design, video dubbing, dictation, transcription and audiobooks, with a local API and MCP server for agents.

View details
86
Repo Health
83
Technical
70
Dependency
Built with
Python 48%
TypeScript 22%
JavaScript 19%
Updated 1 weeks ago
Go
97%
BSD 3

Weaviate

Databases · Search

16,940

Open-source vector database combining semantic search, hybrid queries, RAG, and image search in a single cloud-native system built for production scale.

View details
91
Repo Health
85
Technical
68
Dependency
Built with
Go 97%
Updated 1 weeks ago

Join founders buildingwith open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on Substack
Join 750+ subscribers