i would like to ask two questions regarding deployment:
I am wondering whether it is possible to deploy my own application (including model) either as a pure Python program (with PyTorch model) or as a Docker container with REST interface?
As far as i know this was possible with Algorithmia. So is there a way to continue using Algorithmia?
Thank yo in advance.
Hey @Stony ,
Yes you can deploy your own Python, R, and Java models to MLOps, check out Custom Models for the full details. I'm not a Docker expert but I believe the Portable Prediction Server and MLOps agent would allow you to connect to your model externally for both predictions and monitoring.
If its of interest I had question similar to this on the community about the functionality differences between the Custom Models (brought in by Algorithmia) and the standard DR models here.
@Stony Concurring with my colleague above, yes you can deploy your own custom models to MLOps and the DataRobot University course listed here also covers that in detail:
Hi @Stony - Welcome to the DataRobot Community and thank you for asking your question here. Looks like you received the help you needed? Is that right? If you're all set, can you please select "Accept as Solution" for the answer(s) that helped you? OR, let us know what more you need.
Also, looking forward to hearing more from you as you work through your use cases with DataRobot!