Welcome to SiMa.ai Developer Guide!

SiMa’s software development offering centers around a GStreamer pipeline that can be defined, packaged, and deployed to the SiMa MLA for accelerated inferencing.

At the heart of this ecosystem is Palette, a containerized development environment that serves as the backbone of the SDK. Palette provides powerful tools like ModelSDK for model quantization, testing, and compilation, the MLA Compiler for optimization, Graph Surgery tools for model adaptation and the MPK Tool for packaging and deploying applications. Developers can use Palette independently to build and manage pipelines locally, making it the foundation of SiMa’s development ecosystem. For on-device inference, ModelExecutor provides a simple C++ and Python API that abstracts GStreamer pipeline setup, model loading, and input/output handling, allowing developers to focus on their application logic rather than infrastructure.

To simplify the process further, Edgematic offers a low-code, web-based platform with a visual pipeline interface for cloud-based evaluation, benchmarking, and performance optimization. With Edgematic, customers can explore and evaluate SiMa’s capabilities without the need to purchase or set up hardware, making it easier to get started. By providing an intuitive interface for faster iteration, Edgematic complements Palette, offering a seamless development experience from evaluation to deployment.

Together, Palette, ModelExecutor, and Edgematic deliver a flexible and scalable development experience, enabling developers to choose between comprehensive workflows or rapid on-device inferencing, regardless of the target system architecture.