Skip to content

Hopsworks Documentation

Build, deploy & maintain AI systems#

Features, training data, models and inference on one governed platform.

Your first feature vector, in three steps#

Install the client, connect to a project with an API key, write a feature group and read a feature vector back.

uv venv && source .venv/bin/activate
uv pip install "hopsworks[python]"
hops setup  # opens a browser, picks a project, caches the key
python  # opens the interpreter, the Python lines below go there or in a notebook

import hopsworks


project = hopsworks.login()  # uses the key hops setup cached, else prompts
fs = project.get_feature_store()

uv venv && source .venv/bin/activate
uv pip install "hopsworks[python]"
hops setup  # opens a browser, picks a project, caches the key
hops fg list

import pandas as pd

df = pd.DataFrame(
    {
        "cc_num": [4467360740682089],
        "amount": [12.5],
        "event_time": pd.to_datetime(["2026-01-01T00:00:00Z"]),
    }
)

fg = fs.get_or_create_feature_group(
    name="transactions",
    version=1,
    primary_key=["cc_num"],
    event_time="event_time",
    online_enabled=True,
)
fg.insert(df)

hops fg create transactions --version 1 --primary-key cc_num \
  --features "cc_num:bigint,amount:double" --online
echo '[{"cc_num": 4467360740682089, "amount": 12.5}]' | hops fg insert transactions --version 1

fv = fs.get_or_create_feature_view(
    name="transactions_view",
    version=1,
    query=fg.select_all(),
)
fv.get_feature_vector(entry={"cc_num": 4467360740682089})

hops fv create transactions_view --version 1 --feature-group transactions
hops fv get transactions_view --version 1 --entry "cc_num=4467360740682089"

Next: create a feature group, create a feature view, retrieve feature vectors, or browse the Python API.

Where Hopsworks runs#

  • Use the managed SaaS

    Sign in to the Hopsworks serverless app and create a project. Nothing to install, free tier available.

    Open run.hopsworks.ai ↗

  • Deploy on your cloud or on-prem

    Managed Kubernetes on AWS, Azure or GCP, or an air-gapped data centre. Talk to us to size and install it.

    Contact Hopsworks ↗ · Deployment options

One architecture, three pipelines#

Independent feature, training and inference pipelines, connected by a shared feature store and model registry.

Shared state layer Data sources Snowflakeext BigQueryext Redshiftext S3ext ADLSext GCSext JDBCext Kafka Pipeline FEATURE spark · sql Pipeline TRAINING python Pipeline INFERENCE python Predictions online · batch Feature store features Online RonDB Offline Delta Model registry models Model v1 · v2 Metrics tracked Prediction logs Logged vectors Monitor drift

Find your path#

Developer

Data scientist

Platform engineer

Security engineer

Administrator

Evaluator

By task#

Task Start here
Deploy AWS, Azure, GCP, on-prem, Helm values
Operate Administration, monitoring, alerts, HA and DR, service operations
Troubleshoot Model serving, Python deployments, online ingestion, Jupyter session capacity
Upgrade 3.x to 4.0 migration, Airflow 3 upgrade, Airflow 3 operator notes

APIs