More personal insights on Hopsworks and MLOps. Remember that these are random personal notes in no particular order:
Roles in Machine Learning Projects: For building a professional, robust, maintainable machine learning project you need at least least two people; someone to focus on the actual ML model and other to do MLOps (the way I see it this is data engineering + DevOps).
Hopsworks python client: This connection provides access to projects, feature groups, feature views, model registry and model serving. For using them is wise to have all of the former things in a class object (at least that's the way I'm building my project).
Feature Group Queries as Objects: Hopsworks queries are objects and work for querying from feature groups; read() method is for reading into data frame.
Model Serving: Hopsworks provides model serving capabilities but I haven't had time to mess around with this yet, will probably try next.
Model Registry: Per Hopsworks documentation
It enables developers to publish, test, monitor, govern and share models for collaboration with other teams. The model registry is where developers publish their models during the experimentation phase. The model registry can also be used to share models with the team and stakeholders.
So basically Hopsworks' model registry is for storing ML models through serializable objects like pickle files.
Silverboy
Top comments (0)