Forks indexed
187
Maintained upstream
117
Dormant · 19 archived
70

Deliberately assembled rather than accumulated. 24 repositories covering the production pipeline end to end, from script to delivery. The spine is the Academy Software Foundation stack, the same interchange, colour and image standards used on studio features, plus open production tracking, Fountain screenwriting tooling, and practical post utilities. Every one was checked for provenance, licence, maintenance and install-time behaviour before forking.

lossless-cut Maintained

forked from mifi/lossless-cut

Trims, cuts, and merges video without re-encoding, so nothing degrades and operations complete near-instantly. The correct first tool for handling large camera-original files.

TypeScript★ 43k
moviepy Maintained

forked from Zulko/moviepy

Programmatic video editing in Python: cuts, compositing, effects, and text overlays as code. The scripting layer for anything repetitive or generated.

Python★ 15k
ffsubsync Maintained

forked from smacke/ffsubsync

Automatically synchronizes a subtitle track to its video by aligning against the audio waveform. Fixes drift and offset without manual retiming.

Python★ 7.8k
openshot-qt Maintained

forked from OpenShot/openshot-qt

A mature open-source non-linear video editor with keyframe animation and a full effects set. Useful as a working NLE and as a readable reference for how one is built.

Python★ 6.2k
storyboarder Maintained

forked from wonderunit/storyboarder

Wonder Unit's storyboarding tool: draw a sequence fast, then export to animatic, PDF, or Premiere. Built for getting a shot idea legible quickly rather than for finished art.

JavaScript★ 3.8k
OpenImageIO Maintained

forked from AcademySoftwareFoundation/OpenImageIO

Reads, writes, and processes images across effectively every format a production encounters, behind one API. Includes oiiotool, the workhorse CLI for batch frame manipulation.

C++★ 2.3k
OpenColorIO Maintained

forked from AcademySoftwareFoundation/OpenColorIO

The colour management framework used on essentially every studio feature. Defines the transforms and config that keep an image consistent from camera through VFX to the DCP, including the ACES pipeline.

C++★ 2.1k
OpenTimelineIO Maintained

forked from AcademySoftwareFoundation/OpenTimelineIO

The Academy Software Foundation's interchange format and API for editorial timelines. Moves cuts, transitions, and markers between Premiere, Resolve, Avid, and Flame without conforming by hand, the backbone of any programmatic editorial workflow.

C++★ 2.0k
openexr Maintained

forked from AcademySoftwareFoundation/openexr

The specification and reference implementation of EXR, the high-dynamic-range image format the industry stores frames in. Multi-channel, lossless, and the assumed interchange for VFX plates.

C★ 1.8k
rez Maintained

forked from AcademySoftwareFoundation/rez

The package configuration and deployment system VFX pipelines are built on. Resolves conflicting versions of DCC tools and libraries per-task, which is how a facility runs three Maya versions at once.

Python★ 1.1k
StoryToolkitAI Maintained

forked from octimot/StoryToolkitAI

AI-assisted editing: transcribes footage, makes it searchable by content, and connects to DaVinci Resolve. Aimed squarely at finding the moment in hours of rushes.

Python★ 1.0k
OpenCue Maintained

forked from AcademySoftwareFoundation/OpenCue

Render farm management built at Sony Pictures Imageworks and now ASWF-governed. Job submission, scheduling, and resource allocation for large render workloads.

Python★ 952
xstudio Maintained

forked from AcademySoftwareFoundation/xstudio

DNEG's playback and review application, donated to the ASWF. Handles dailies and review sessions with annotation and colour-accurate playback at production frame rates.

C++★ 766
kitsu Maintained

forked from cgwire/kitsu

CGWire's production tracking platform for animation and VFX: shots, assets, tasks, and review, with a real scheduling model. The credible open alternative to ShotGrid.

Vue★ 685
Fountain Maintained

forked from nyousefi/Fountain

The original reference implementation of the Fountain screenplay markup language. The parser to read if you are building anything that consumes screenplays.

Objective-C★ 673
betterfountain Maintained

forked from piersdeseilligny/betterfountain

Screenwriting inside VS Code using the Fountain plain-text format: live preview, industry-standard PDF export, and production statistics. Keeps scripts in git alongside everything else.

JavaScript★ 447
fountain-mode Maintained

forked from rnkn/fountain-mode

An Emacs major mode for writing screenplays in Fountain, with outlining, navigation, and export.

Emacs Lisp★ 440
scrite Maintained

forked from teriflix/scrite

A cross-platform desktop screenwriting application with structure and scene management built in, not just formatting.

C++★ 365
afterwriting-labs Maintained

forked from ifrost/afterwriting-labs

Post-processing for Fountain screenplays. PDF generation, statistics, and formatting control beyond what most editors expose.

JavaScript★ 276
shotmanager Maintained

forked from ubisoft/shotmanager

Ubisoft's Blender add-on introducing a real shot entity into a scene, with a shot-based editing and previz workflow. The practical route to layout and previs inside Blender.

Python★ 215
zou Maintained

forked from cgwire/zou

The API server behind Kitsu. Worth having separately if you want production data in your own tooling rather than through the web client.

Python★ 195
gazu Maintained

forked from cgwire/gazu

The Python client for the Kitsu API, the practical way to script against production tracking data from inside a DCC or a pipeline tool.

Python★ 96

Foundation and generative models themselves: speech, vision-language, music, 3D, and the frontier open-weight releases.

whisper Maintained

forked from openai/whisper

OpenAI's robust multilingual speech recognition model. The default choice for transcription; trained with large-scale weak supervision.

Python★ 107k
VibeVoice Maintained

forked from microsoft/VibeVoice

Microsoft's open frontier voice model for long-form, multi-speaker speech synthesis. One of the strongest open TTS releases available.

Python★ 53k
CLIP Maintained

forked from openai/CLIP

OpenAI's contrastive image-text model. The backbone behind most modern multimodal retrieval, captioning, and generation systems.

Jupyter Notebook★ 34k
GLM-5 Maintained

forked from zai-org/GLM-5

Z.ai's frontier open-weight model release, positioned around agentic engineering rather than chat. Useful as a reference for what an open frontier model card and eval suite look like.

★ 6.9k
point-e Maintained

forked from openai/point-e

OpenAI's point-cloud diffusion system for fast text-to-3D synthesis. Directly relevant to any image-to-3D pipeline.

Python★ 6.9k
SCAIL-2 Maintained

forked from zai-org/SCAIL-2

Z.ai's controlled character-animation model: unifies pose, motion, and identity control through end-to-end in-context conditioning.

Python★ 1.1k
VibeVoiceFusion Maintained

forked from zhao-kun/VibeVoiceFusion

Full-stack web system wrapped around VibeVoice: multi-speaker generation, LoRA fine-tuning, batch jobs, and VRAM management.

Python★ 489
Vision2Web Maintained

forked from zai-org/Vision2Web

Z.ai research turning screenshots and visual designs into working web code, the design-to-code direction as a model problem.

Python★ 67

The agent layer: Claude and OpenAI tooling, RAG demos, eval harnesses, and agent frameworks you can lift patterns from.

openai-cookbook Maintained

forked from openai/openai-cookbook

OpenAI's canonical example collection: prompting, embeddings, function calling, evals. The single best API-patterns reference.

Jupyter Notebook★ 75k
career-ops Maintained

forked from santifer/career-ops

A larger take on the same idea: 14 skill modes, a Go dashboard, PDF generation, and batch processing over job listings.

JavaScript★ 64k
CLIProxyAPI Maintained

forked from router-for-me/CLIProxyAPI

Wraps Claude Code, Codex, Gemini, and Grok behind one OpenAI-compatible API surface. A practical reference for provider-abstraction layers.

Go★ 47k
ai-job-search Maintained

forked from MadsLorentzen/ai-job-search

Claude Code-based job application framework: fill in a profile and let the agent evaluate roles, tailor CVs, and draft cover letters.

TypeScript★ 31k
openai-python Maintained

forked from openai/openai-python

The official OpenAI Python SDK: useful for reading how a mature LLM client handles streaming, retries, and typed responses.

Python★ 31k
evals Maintained

forked from openai/evals

OpenAI's framework and open registry for evaluating LLMs and LLM systems. Relevant to any accuracy-gating harness.

Python★ 19k
llm-zoomcamp Maintained

forked from DataTalksClub/llm-zoomcamp

DataTalksClub's free 10-week course on building real LLM applications end to end, including RAG and monitoring.

Jupyter Notebook★ 7.0k
image-blaster Maintained

forked from neilsonnn/image-blaster

An image-to-world skillset packaged as a Claude skill, a working example of the skill-authoring format.

TypeScript★ 4.8k
AgenticOS Maintained

forked from ChainGPT-org/AgenticOS

ChainGPT's TypeScript AI agent for Twitter/X built on Hono and Bun: autonomous tweet generation and engagement loops.

TypeScript★ 30

The training stacks and pretrained model collections everything else is built on top of.

models Maintained

forked from tensorflow/models

TensorFlow's official model garden: reference implementations across vision, NLP, and recommendation, maintained alongside the framework.

Python★ 78k
keras Maintained

forked from keras-team/keras

The Keras high-level deep learning API, now multi-backend across TensorFlow, JAX, and PyTorch.

Python★ 64k
tfjs Maintained

forked from tensorflow/tfjs

TensorFlow.js. WebGL-accelerated training and inference directly in the browser.

TypeScript★ 19k
tfjs-models Maintained

forked from tensorflow/tfjs-models

Ready-to-use pretrained models for TensorFlow.js: pose estimation, object detection, face mesh, text toxicity.

TypeScript★ 15k
tensorboard Maintained

forked from tensorflow/tensorboard

TensorFlow's visualization toolkit for scalars, graphs, embeddings, and profiling runs.

TypeScript★ 7.2k
docs Maintained

forked from tensorflow/docs

The TensorFlow documentation source, including the tutorial notebooks published on tensorflow.org.

Jupyter Notebook★ 6.3k
build Maintained

forked from tensorflow/build

TensorFlow's build tooling. Docker images, CI configuration, and the release pipeline for the framework itself.

Shell★ 306

Text processing, translation, topic modeling, and word-embedding tooling.

spaCy Maintained

forked from explosion/spaCy

Industrial-strength NLP in Python: tokenization, NER, dependency parsing, and pipelines built for production throughput.

Python★ 34k
BERTopic Maintained

forked from MaartenGr/BERTopic

Topic modeling that combines transformer embeddings with c-TF-IDF to produce interpretable, labeled topic clusters.

Python★ 7.8k
AIND-NLP Maintained

forked from udacity/AIND-NLP

Udacity AI Nanodegree NLP exercises: text processing, feature extraction, and sequence modeling.

Jupyter Notebook★ 241

Image, video, and 3D asset work, including the 2D→3D conversion pipeline forks.

particles.js Maintained

forked from VincentGarreau/particles.js

A lightweight JavaScript library for particle backgrounds. The standard reference for canvas-based ambient effects.

JavaScript★ 30k
doctr Maintained

forked from mindee/doctr

Mindee's OCR library: end-to-end text detection and recognition over documents, with both TensorFlow and PyTorch backends.

Python★ 6.2k
deepgaze Maintained

forked from mpatacchiola/deepgaze

Computer vision library for human-computer interaction: head pose estimation, gaze direction, and face tracking with CNNs.

Python★ 1.9k
ovc Maintained

forked from osrf/ovc

The Open Vision Computer from OSRF: open hardware and firmware for a stereo camera and IMU module used in robotics.

C★ 209

Hyperparameter search, explainability, data validation, and serving, the unglamorous half of shipping ML.

shap Maintained

forked from shap/shap

Game-theoretic feature attribution for any model. The standard tool for explaining individual predictions.

Jupyter Notebook★ 26k
modin Maintained

forked from modin-project/modin

Scales Pandas workloads across cores or a cluster by changing a single import line.

Python★ 10k
hyperopt Maintained

forked from hyperopt/hyperopt

Distributed asynchronous hyperparameter optimization in Python. TPE and random search over awkward, conditional search spaces.

Python★ 7.6k
hyperas Maintained

forked from maxpumperla/hyperas

A thin Keras wrapper over Hyperopt that makes hyperparameter search readable inside a normal model definition.

Python★ 2.2k
talos Maintained

forked from autonomio/talos

Hyperparameter experiment management for TensorFlow and Keras: scans, evaluation, and deployment of the winning configuration.

Python★ 1.6k
python-deequ Maintained

forked from awslabs/python-deequ

AWS's Python API for Deequ: data quality constraint checks running at scale on Spark.

Jupyter Notebook★ 826
datatableton Maintained

forked from vopani/datatableton

100 graded exercises for learning Python's datatable, a fast alternative to Pandas for large frames.

Jupyter Notebook★ 273

GCP, AWS, and Azure SDKs and samples, Kubernetes and Terraform patterns, and small networking servers.

Real analyses and public-facing dashboards built on civic, social, and web data.

Trading strategy code and fraud-detection notebooks.

Coursework, specializations, curated curricula, and study notes, the largest slice of the archive.

dlaicourse Maintained

forked from lmoroney/dlaicourse

Laurence Moroney's notebooks for the DeepLearning.AI TensorFlow courses, the official exercise source.

Jupyter Notebook★ 5.6k
bookclub Maintained

forked from SanDiegoMachineLearning/bookclub

Notes and links from the San Diego Machine Learning book club, a running reading list with discussion summaries.

Jupyter Notebook★ 781
Talks Maintained

forked from georgezoto/Talks

An index of repositories backing conference talks, meetups, and presentations.

HTML★ 0

Core systems source, editors, regex engines, and curated link collections.

linux Maintained

forked from torvalds/linux

The Linux kernel source tree. Fork of torvalds/linux, the reference for anything at the OS layer.

C★ 243k
word_cloud Maintained

forked from amueller/word_cloud

The standard Python word cloud generator, with masking, coloring, and stopword control.

Python★ 11k
re2 Maintained

forked from google/re2

Google's linear-time regular expression engine. Safe against catastrophic backtracking, unlike PCRE-style engines.

C++★ 9.8k
api-mega-list Maintained

forked from cporter202/API-mega-list

A large curated index of public APIs across every category, the fastest starting point when scoping an integration.

JavaScript★ 7.5k

Upstream is formally archived, or has had no commit in five years or more. These are pulled out of their topic sections so the rest of the index reflects what is still maintained, each card keeps the topic it came from. Archived is a decision someone made; dormant is just silence, and a few of these (BERT, tensor2tensor, the CS 229 sheets) are still the canonical reference for their subject.

smart-city Dormant

forked from georgezoto/smart-city

The OCTO Smart City inventory, a catalog of municipal sensor and connectivity projects.

JavaScript★ 0quiet 10y
phd-thesis Dormant

forked from glouppe/phd-thesis

Gilles Louppe's thesis β€œUnderstanding Random Forests”, the definitive theoretical treatment of tree ensembles.

TeX★ 529quiet 10y
attention-nmt Dormant

forked from pemywei/attention-nmt

A compact, readable attention-based encoder-decoder for machine translation. Good for understanding the mechanism without framework noise.

Python★ 46quiet 9y
mlsh Dormant

forked from georgezoto/mlsh

Meta-Learning Shared Hierarchies: reinforcement learning agents that learn reusable sub-policies across task distributions.

Python★ 0quiet 9y
Tableau-A-Z Dormant

forked from georgezoto/Tableau-A-Z

Course material for the Tableau A-Z hands-on training: dashboards, joins, and calculated fields.

★ 3quiet 8y
tfjs-vis Archived

forked from tensorflow/tfjs-vis

In-browser visualization utilities for TensorFlow.js: training curves, confusion matrices, and layer inspection.

TypeScript★ 372quiet 7y
teach_DL Dormant

forked from miroenev/teach_DL

Teaching materials and notebooks for introducing deep learning with GPU-accelerated examples.

Jupyter Notebook★ 35quiet 7y
VisualizeMNIST Dormant

forked from okdalto/VisualizeMNIST

A real-time Processing visualization of a neural network recognizing hand-drawn digits. Excellent for explaining inference visually.

Processing★ 598quiet 7y
fashion-mnist Dormant

forked from georgezoto/fashion-mnist

A drop-in MNIST replacement using clothing images, the standard harder-than-MNIST benchmark for quick model sanity checks.

★ 0quiet 7y
scibert Dormant

forked from georgezoto/scibert

BERT pretrained on scientific literature, for domain-specific NLP over papers and technical text.

★ 0quiet 6y
fairseq Dormant

forked from vineetk1/fairseq

A transaction-bot built on Meta's fairseq toolkit using neural attention-based encoder-decoder models.

Python★ 4quiet 6y
VMMRdb Dormant

forked from faezetta/VMMRdb

Vehicle Make and Model Recognition dataset: real-world images across thousands of make/model/year classes.

Lua★ 175quiet 6y
oso Dormant

forked from georgezoto/oso

An open-source policy engine for authorization that embeds directly in your application, replacing scattered permission checks.

★ 0quiet 6y
brackets Archived

forked from adobe/brackets

Adobe's open-source web code editor, written in JavaScript, HTML, and CSS. Now archived but a landmark editor codebase.

JavaScript★ 33kquiet 4y
nmt Archived

forked from tensorflow/nmt

TensorFlow's official neural machine translation tutorial: seq2seq with attention, explained end to end.

Python★ 6.5kquiet 4y
docker-lambda Archived

forked from lambci/docker-lambda

Docker images that replicate the live AWS Lambda runtime, so functions can be tested locally with real behavior.

C#★ 5.8kquiet 4y
tensor2tensor Archived

forked from tensorflow/tensor2tensor

The earlier Google library of models and datasets that introduced many transformer training recipes. Archived but historically important.

Python★ 17kquiet 3y
jukebox Archived

forked from openai/jukebox

OpenAI's raw-audio music generation model: generates song audio with vocals conditioned on genre, artist, and lyrics.

Python★ 8.0kquiet 2y
bert Archived

forked from google-research/bert

Google Research's original BERT release. TensorFlow code plus pretrained checkpoints. The paper that started the pretraining era.

Python★ 40kquiet 2y
trax Archived

forked from google/trax

Google Brain's clear-code deep learning library, built for readable research implementations of transformers and RL.

Python★ 8.3kquiet 1y