2 minute read
From Pitch to Acceptance: OAAX Joins LF AI & Data
Written by Chief Data Scientist, Maurits Kaptein and AI/Machine Learning Engineer, Ayoub Assis.
On January 16th, the OAAX team had the privilege of presenting our vision to the LF AI & Data Technical Advisory Council. This presentation was an important opportunity to showcase the transformative potential of the Open AI Accelerator eXchange (OAAX), an open standard designed to simplify the development and deployment of edge AI applications.
Today, we are proud to announce that OAAX has been officially accepted as a Sandbox Project under the Linux Foundation AI & Data! This milestone highlights our commitment to fostering open collaboration and addressing the challenges of deploying AI in today’s fragmented hardware landscape.
Why LF AI & Data? A Community-Driven Mission
Joining LF AI & Data is about more than just governance—it’s about community. By becoming a part of this global initiative, OAAX aligns with a shared mission to advance AI through open standards and collaborative innovation. The foundation provides invaluable resources, such as neutral governance, technical infrastructure, and access to a vibrant developer network.
This partnership also enables us to focus on building something truly impactful: a unified, standardized solution for AI deployment that benefits everyone in the ecosystem.
Our Commitment at Network Optix
At Network Optix, our mission is to empower developers and organizations to build and deploy cutting-edge Video AI applications with ease. This commitment drives our support for OAAX and adherence to the OAAX standard, allowing us to create software that works seamlessly across platforms and with virtually any accelerator.
By making this a community-driven effort, we give back our developments to the broader AI ecosystem. This approach not only simplifies the deployment of production-ready Video AI applications but also accelerates the maturity of tools and technologies available to everyone. In turn, this strengthens the foundation for innovation and collaboration in AI.
The Power of Standardization: A Familiar Analogy
Imagine driving a car. No matter the brand or model, you can rely on standardized controls—steering wheel, pedals, and dashboard. This consistency means you can step into any car and drive without needing specialized training.
OAAX applies the same principle to AI accelerators. Currently, deploying AI models often requires navigating proprietary toolchains and unique runtime environments. With OAAX, we’re creating a universal “dashboard” for AI accelerators, enabling developers to deploy their models effortlessly across platforms.
This isn’t just about convenience. It’s about enabling innovation, reducing costs, and ensuring that anyone can build and deploy AI applications without being locked into specific hardware ecosystems.
How OAAX Works
OAAX is driven by two standardized processes:
- The Toolchain: Provided by the community or XPU manufacturers, a toolchain takes a trained model in ONNX format (potentially with additional resources) as input, and outputs a log file and a file that can be loaded by the associated runtime. Toolchains can run on any machine with sufficient resources—be it a server, laptop, or other devices.
- The XPU-Specific Runtime: A standardized library to load the output of the runtime (the model that has been optimized for the XPU) and generate inferences. The runtime is run on the edge device.
Like car dashboards that standardize controls but still allow individual car manufacturers to showcase their engines and driving experience, OAAX makes using any XPU remarkably easy, while giving XPU manufacturers the flexibility to highlight the unique capabilities of their hardware.
What’s Next for OAAX?
Acceptance into LF AI & Data as a Sandbox project is just the beginning. This partnership provides us with the resources, infrastructure, and community support needed to refine and scale OAAX. Together, we’re laying the groundwork for a unified standard that makes deploying AI easier, more accessible, and more reliable.
We invite developers, organizations, and hardware manufacturers to join us in this effort. By working together, we can make AI deployment as seamless and intuitive as driving a car—an experience anyone can access and benefit from.
Learn more about OAAX on our GitHub repository. Let’s build the future of AI deployment—together!
Additional Resources
- Demystifying OAAX: A Closer Look at Open AI Accelerator eXchange
- Introducing the Open AI Accelerator eXchange: Streamlining Edge AI Deployment