In today's connected world, consumer devices are generating vast amounts of data, and there's a growing need to perform AI workloads locally to ensure data privacy. This approach also enables real-time processing, reduces server costs, and unlocks new use cases that were previously impossible.
With numerous open-source projects and models available today, this is an ideal opportunity to explore the theme of on-device AI for consumer use-cases.
In the spirit of a hackathon, you're encouraged to use any open-source tools and models (make sure to comply with their respective licenses). Our team is here to provide guidance, bounce off ideas, and offer expertise in specific technologies as needed. Also, contributions to our repos (e.g., ExecuTorch, torchchat, HuggingFace transformers etc) are highly encouraged, if doing so helps to fit your use-cases better.
To get you started, we've compiled a list of potential project ideas. However, feel free to think outside the box and explore new concepts – the possibilities are endless!
In general, there are a few technical challenges that we see for on-device AI to achieve “mainstream”:
During the hackathon, it would be great to think about how your project solves these challenges.
⚙️ To tackle these challenges and more, make sure to bring your own hardware / microprocessing boards if needed.