Open Source

Building AI in the Open

We believe in transparency. Our research code, tools, and datasets are available for the community to use, study, and improve.

Featured Repositories

Our most impactful open-source projects

Production-ready inference framework with built-in safety guardrails, output filtering, and comprehensive logging for responsible AI deployment.

Python PyTorch CUDA
3.2K 456 forks Updated 2 days ago

Interactive cost-quality-speed tradeoff analyzer based on our Tradeoff Selector™ methodology. Helps teams make informed decisions about AI system configurations.

TypeScript React D3.js
892 134 forks Updated 1 week ago

Comprehensive benchmarking toolkit for evaluating fairness and bias in ML models across multiple protected attributes and use cases.

Python PyTorch scikit-learn
1.8K 289 forks Updated 5 days ago

Interpretability toolkit providing SHAP, LIME, attention visualization, and concept-based explanations in a unified, user-friendly API.

Python PyTorch TensorFlow
2.4K 367 forks Updated 3 days ago

Document integrity verification using cryptographic hashing and ML-based tamper detection. Detects manipulated PDFs with 99.7% accuracy.

Rust Python bindings WebAssembly
1.1K 156 forks Updated 1 day ago

Calibrated uncertainty estimation for deep learning models using conformal prediction, ensembles, and Bayesian methods.

Python PyTorch JAX
967 112 forks Updated 4 days ago

Open Datasets

Research datasets freely available for academic and commercial use

TERA-BIAS-21

250K annotated examples for bias detection across 7 protected categories in NLP tasks.

📊 250,000 samples 📝 CC BY 4.0 🔄 v2.1

DocTamper-Bench

50K pristine and 50K tampered documents for training document integrity models.

📊 100,000 documents 📝 Apache 2.0 🔄 v1.3

MedSafe-Eval

Evaluation dataset for healthcare AI safety with expert physician annotations.

📊 15,000 cases 📝 Research Only 🔄 v1.0

Uncertainty-Calib

Multi-domain benchmark for calibration and uncertainty quantification methods.

📊 500,000 samples 📝 MIT 🔄 v2.0

How to Contribute

Join our community of contributors building trustworthy AI

1

Find an Issue

Browse our repositories for issues labeled "good first issue" or "help wanted" to find a task that matches your skills.

2

Fork & Develop

Fork the repository, create a feature branch, and implement your changes following our contribution guidelines.

3

Submit PR

Open a pull request with a clear description. Our maintainers will review and provide feedback within 48 hours.

Build with Us

Whether you're a researcher, engineer, or student, there's a place for you in our open-source community. Let's build responsible AI together.