hopsworks

What it is and what it does

Hopsworks is a Python client library for the Hopsworks platform, an MLOps and DataOps system centered on a feature store. It lets you connect to a Hopsworks cluster (cloud-hosted or self-managed) to manage features, training datasets, ML models, and real-time serving. The library supports two operational modes: Spark mode for large-scale data engineering jobs that write features into the store, and Python mode for data science workflows that read features, build training datasets, and deploy models for inference.

The package wraps three main APIs: the Feature Store API (create feature groups, insert dataframes, build training datasets, serve real-time feature vectors), the Model Registry API (register and version ML models), and the Model Serving API (deploy and manage model endpoints). It depends on 24 runtime packages including pandas, numpy, boto3, grpcio, and opensearch-py to handle data processing, cloud storage, gRPC communication, and search operations. Most workflows begin with `hopsworks.login()` to authenticate against your cluster, then branch into feature store, model registry, or serving operations.

Use it for

• Build and manage a centralized feature repository: create feature groups from dataframes, version them, and share features across multiple ML projects.
• Generate training datasets: query feature groups, join them, and export as CSV, Parquet, or TFRecord for model training pipelines.
• Serve real-time predictions: fetch online feature vectors on demand (combining stored features with request-time context) to feed into deployed models.
• Register and deploy ML models: version TensorFlow, PyTorch, or other models in the registry, then deploy them as HTTP endpoints for inference.
• Integrate with cloud storage: use storage connectors to ingest data from S3, GCS, or other sources into the feature store.
• Monitor and manage Hopsworks infrastructure: on self-hosted clusters, manage environments, Kafka topics, and job executions via the API.

Worth the install?

AI-flagged interpretation of the facts on this page. Verify before relying on it.

Install

hopsworks on PyPI

Before you install

Low install friction with a pure-wheel distribution. Active maintenance with a release 9 days old; last commit 2026-08-14. Supports Python 3.10–3.13.

Requires a Hopsworks cluster (SaaS or self-hosted) and an API key to authenticate; Python mode works with just a Python interpreter but still needs network access to the cluster.

License in practice

Apache-2.0 (permissive): you can use, modify, and distribute this package freely in commercial and private projects, provided you include a copy of the license and state significant changes.

Quickstart

pip install "hopsworks[python]"

import hopsworks
project = hopsworks.login()
fs = project.get_feature_store()
fg = fs.get_or_create_feature_group(name="my_features", version=1)
fg.insert(my_dataframe)

Package facts

Evidence: hopsworks-5.0.4-py3-none-any.whl

Tags

More Utilities packages

See also kubeflow · arthur-client · ibm-watson-machine-learning · sagemaker-feature-store-pyspark-3.1 · skypilot-nightly · snowflake-ml-python · sagemaker · tinker · IxNetwork · kumoai