Introducing Nillion’s Nada AI Developer Toolkit






Tech Updates

We’re excited to announce the release of two groundbreaking open-source libraries: Nada Numpy and Nada AI. These tools empower developers to build privacy-preserving AI systems on the Nillion Network, addressing one of the most pressing issues in the rapidly advancing field of artificial intelligence: data privacy.

Nillion’s Commitment to Privacy-Preserving AI

As AI continues to disrupt various industries—from healthcare and finance to transportation and entertainment—concerns about data privacy have intensified. In response, Nillion is focused on developing novel privacy-preserving technologies that enable the creation of effective private AI systems.

Our approach is rooted in familiarity and accessibility. We’ve built upon the foundation of widely-used, open-source AI tools like NumPy, PyTorch, Scikit-Learn, Matplotlib, and TensorFlow. These tools have become the gold standard in the AI community and we’re committed to providing developers with what they know and love while prioritizing data privacy.

Introducing Nada Numpy and Nada AI

With this foundation in mind, let’s explore our two new libraries: Nada Numpy and Nada AI. As their names suggest, these libraries are adaptations of popular Python tools, tailored to our proprietary language, Nada DSL. Designed by AI developers for AI developers, Nada Numpy and Nada AI seamlessly integrate with existing workflows, bridging the gap between familiar tools and privacy-preserving computation.

Nada Numpy

Nada Numpy offers a subset of NumPy directives for use with Nada, aiming to bring the user-friendly NumPy experience to the Nada DSL environment. Our goal with Nada Numpy is to allow users to seamlessly reuse most of their existing NumPy code while interacting with Nada DSL.

Key features include:

  • Preservation of beloved NumPy features and functions
  • Support for complex functionality within the evolving Nada DSL
  • Efficient array manipulation within Nada DSL: including array operations, broadcasting and reshaping

Here’s a side-by-side comparison of Nada Numpy and regular NumPy syntax:


Nada AI

Nada AI is an all-in-one solution designed to seamlessly integrate AI models from other frameworks into the Nillion ecosystem. 

Key features include:

  • Integration of AI models in Nada DSL
  • Uploading programs and storing models on the Nillion Network
  • Privacy-preserving inference process
  • Comprehensive interface with bridges to PyTorch, Scikit-Learn linear models, and Prophet Time Series

What sets our solution apart is the privacy-preserving inference process. Using Nillion’s proprietary secure multi-party computation technology, we enable computations on encrypted data without revealing the underlying information. This means that during the inference process, neither the network or users learn anything about each other’s information, while still achieving accurate predictions based on the initial input.

Building a Privacy-Preserving AI Future

The release of Nada AI and Nada Numpy is just the beginning of our journey to provide developers with the tools needed to create production-grade, privacy-preserving AI applications. We’re committed to continuously improving these libraries, incorporating valuable feedback from our growing community of developers and pushing the boundaries of what’s possible in confidential AI.

Whether you’re a seasoned AI developer or just starting your journey, we’d love to hear from you. Your feedback is crucial in helping us refine these tools. Dive into our official Nada AI Documentation and explore our Github Repository. Together, we can build a future where AI and privacy go hand in hand.

About Nillion

Nillion is humanity’s first Blind Computer. It is powered by a decentralized network of nodes that enables “Blind Computation” through the coordination and orchestration of privacy enhancing technologies (PETs) such as multi-party computation (MPC), fully homomorphic encryption (FHE) and zero-knowledge proofs (ZKP). Nillion believes Blind Computation will become the internet’s base layer for all private data as PETs continue to mature. Nillion has attracted a notable initial cohort of Blind Computation builders across AI, DeFi, medical data, custody, wallets, global identity, messaging, and more.

Learn more by visiting the Nillion website or following us on TwitterTelegram or Discord.


[1] Meskó, B., & Topol, E. J. (2023). The imperative for regulatory oversight of large language models (or generative AI) in healthcare. NPJ digital medicine, 6(1), 120.

[2] Cao, L. (2022). Ai in finance: challenges, techniques, and opportunities. ACM Computing Surveys (CSUR), 55(3), 1-38.

[3] Geyer, M., Bar-Tal, O., Bagon, S., & Dekel, T. (2023). Tokenflow: Consistent diffusion features for consistent video editing. arXiv preprint arXiv:2307.10373.