Private machine learning done right (Ep. 207)
Data Science at Home
00:00
Using Onyx to Deploy a Regular Model
Onyx is a kind of standard that has been evolving recently to train on deploying a models. It has been pushed around by Microsoft and Facebook but also Nvidia. And so in our case, we have taken track, which is an inference engine in rest that was initially made for embedded inference. And then blind AI is the middle layer that is able to load the, the model in security and the enclave, serve the model to the client and so on. We're working on also finding providing this abstraction layer, this third layer to make it even more easy to manage users, key management and etc.
Transcript
Play full episode
