open source helper/framework based on Python for speedup delivery of ML/AI to the production (support MLOps with featurestore concept), main focus on kubernetes solution. Use this tag for questions about platform MLRun (e.g. MLRun CE, MLRun paid version) or libraries from this solution (e.g. v3io, v3io stream, storey).
MLRun is an open source helper/framework based on Python for speedup delivery from ML/AI experiment phase to production with focus on minimalizing development effort (agnostic approach without focus on a specific AI/ML topic).
It supports MLOps with featurestore concept, including continuous training and continuous monitoring also. It is based on K8s with ability to support scalability of AI/ML in on-prem and cloud (support of main three major cloud providers).
Sample of python code:
import mlrun
import mlrun.feature_store as fstore
# create project 'test'
project = mlrun.get_or_create_project("test", context="./")
# add feature set (part of feature store)
transaction_set = fstore.FeatureSet("transaction", entities[fstore.Entity("trid")])
# ingest data
fstore.ingest(transaction_set, bank_trn, infer_options=fstore.InferOptions.default())
More information
