Bring Your Own Model — A Local, Privacy-Focused Machine Learning Tool
BYOM is a standalone system for training, running, and reusing machine learning models entirely on your own computer.
This video demonstrates BYOM in action — training and classifying models locally on your machine, with real-time outputs sent via OSC.
BYOM puts power back in the hands of the user. All computation happens locally — no data is sent to cloud servers or large corporations. By keeping AI training and inference on your own machine, BYOM prioritizes privacy, control, and personal empowerment.
currently Users can train models for image classification , audio classification, and pose recognition, then send real-time classification outputs via OSC (Open Sound Control) to creative applications, performances, and interactive installations.
BYOM emphasizes hands-on learning and experimentation. Users teach models by example, immediately seeing how the system responds, without relying on cloud platforms or proprietary AI services. Users can also edit the laers of the model depending on how much capacity for training on their hardware is.
By operating entirely locally, BYOM fosters understanding, transparency, and creative freedom. It empowers artists, researchers, and educators to explore machine learning safely and responsibly.
BYOM runs completely offline and requires no internet connection. Users can train, evaluate, and deploy models within the same interface. Real-time classification is streamed via OSC, making it compatible with software such as Max/MSP, TouchDesigner, SuperCollider, Unity, and custom toolchains.
The project explores how local AI interfaces can democratize machine learning, giving people the ability to experiment and innovate without relying on large corporate platforms or surrendering their data.