
Integration guides
Hosted API
Get started in minutes. Create an account, grab your API key, and run predictions from anywhere. Noinfrastructure to set up — we handle the GPUs, the scaling, and the availability.
from neuralk import Seldon
clf = Seldon(api_key="nk_live_xxx")
clf.fit(X_train, y_train)
predictions = clf.predict(X_test)On-premise deployment
For organizations that require full data sovereignty, Neuralk deploys on your infrastructure. Your data never leaves your perimeter. Same SDK, same developer experience — just point it to your own endpoint. Compatible with Kubernetes, private cloud, and bare-metal environments.
from neuralk import Seldon
clf = Seldon("https://seldon.internal.your-company.com)")
clf.fit(X_train, y_train)
predictions = clf.predict(X_test)Connect your data
Neuralk works with the data tools you already use. Load your datasets from any source, pass them to the classifier — we handle the rest.
DataFrames

Pandas

Polars
Arrays

NumPy
File formats

Parquet

CSV

Excel
Cloud storage

Amazon S3

Google Cloud Storage

Azure Blob Storage
Data warehouses

Snowflake

BigQuery

Redshift

Databriks
Databases

PostgreSQL

MySQL

Any ODBC/JDBC source
import pandas as pd
from neuralk import Seldon
# Load from wherever your data lives
df_train = pd.read_parquet("s3://my-bucket/train.parquet")
df_test = pd.read_parquet("s3://my-bucket/test.parquet")
clf = Seldon(api_key="nk_live_xxx")
clf.fit(df_train[features], df_train[target])
predictions = clf.predict(df_test[features])Scikit-learn compatible
Neuralk implements the scikit-learn estimator interface. It plugs directly into your existing ML pipelines — cross-validation, grid search, pipelines, and model selection all work out of the box.
from sklearn.model_selection import cross_val_score
from neuralk import Seldon
clf = Seldon(api_key="nk_live_xxx")
scores = cross_val_score(clf, X, y, cv=5)