The Zant Project

Project Overview

We are builing an open-source SDK designed to simplify and cost-effectively deploy machine learning models on embedded and edge devices. With Zant, developers wiil be able to easily optimize and deploy models on a wide range of hardware, minimizing the need for complex reimplementation when switching platforms.
The first release of Zant will be a static library that takes an ML model as input and produces an optimized, device-specific executable. Built primarily in the Zig programming language, Zant leverages two powerful features of Zig:
Cross-Compilation: Zant enables seamless code portability, allowing ML models to run on different device architectures with minimal adjustments. This ensures flexibility and saves development time, especially in resource-constrained environments.
C-Compatibility: As C is the standard language for embedded applications, Zant’s compatibility with C allows it to integrate smoothly with essential components like the Hardware Abstraction Layer (HAL), which provides a consistent interface for hardware interactions.
With Zant, deploying ML models to embedded and edge devices becomes more efficient, flexible, and accessible.

Why this matter:

1. Many microcontrollers (e.g., ATMEGA, TI Sitara) lack robust deep learning libraries.
2. No open-source solution exists for end-to-end NN optimization and deployment.
3. Inspired by cutting-edge research (e.g., MIT Han Lab), we leverage state-of-the-art optimization techniques.
4. Built for flexibility to adapt to new hardware without codebase changes.

Key Features:

• Lightweight & High-Performance: Small footprint, ideal for constrained environments.
• Hardware Acceleration: Supports SIMD, GPU offloading for enhanced efficiency.
• Efficient Memory Management: Uses memory pooling and static allocation.
• Optimized Algorithms: Includes quantization, model pruning for embedded AI.
• Cross-Platform: Compatible with ARM Cortex-M, RISC-V, and more.

How to Contribute

We welcome contributors of all experience levels and backgrounds. A strong desire to learn and passion for the project are required.

Note that we wish to build a company around the Zant project; we will only hire former contributors to the project.

Fill this form to join the project

FAQ