Console’s review
· Reviewed by David Mytton
What we like
Build & deploy models from Jupyter notebooks. Provides an inference endpoint accessible via a REST JSON API and from within data warehouses e.g. from SQL, Redshift, or Snowflake. Supports custom environments, packages, tests. Can also run training jobs from a notebook or from Git. Includes logs, monitoring & version control.
Tradeoffs
Everything is based around Python & Notebooks - this is the standard ML stack, but worth being aware of if you’re using something else.