Chapkit

Build production-ready ML services with train/predict workflows, artifact storage, config management, and job scheduling - all in a few lines of code.

Quick Start: ML Service

from geojson_pydantic import FeatureCollection

from chapkit import BaseConfig
from chapkit.api import MLServiceBuilder, MLServiceInfo
from chapkit.artifact import ArtifactHierarchy
from chapkit.data import DataFrame
from chapkit.ml import FunctionalModelRunner


class MyMLConfig(BaseConfig):
    """Configuration for your ML model."""


async def train_model(
    config: MyMLConfig,
    data: DataFrame,
    geo: FeatureCollection | None = None,
) -> dict:
    """Train your model - returns trained model object."""
    df = data.to_pandas()
    # Your training logic here - example using sklearn:
    # from sklearn.linear_model import LinearRegression
    # model = LinearRegression()
    # model.fit(df[["feature1", "feature2"]], df["target"])
    return {"trained": True}


async def predict(
    config: MyMLConfig,
    model: dict,
    historic: DataFrame,
    future: DataFrame,
    geo: FeatureCollection | None = None,
) -> DataFrame:
    """Make predictions using the trained model."""
    future_df = future.to_pandas()
    # Your prediction logic here
    future_df["sample_0"] = 0.0  # Replace with actual predictions
    return DataFrame.from_pandas(future_df)


app = (
    MLServiceBuilder(
        info=MLServiceInfo(display_name="Disease Prediction Service"),
        config_schema=MyMLConfig,
        hierarchy=ArtifactHierarchy(
            name="ml",
            level_labels={0: "ml_training_workspace", 1: "ml_prediction"},
        ),
        runner=FunctionalModelRunner(on_train=train_model, on_predict=predict),
    )
    .with_monitoring()  # Optional: Add Prometheus metrics
    .build()
)

What you get: - POST /api/v1/ml/train - Train models with versioning - POST /api/v1/ml/predict - Make predictions - GET /api/v1/configs - Manage model configurations - GET /api/v1/artifacts - Browse trained models and predictions - GET /api/v1/jobs - Monitor training/prediction jobs - GET /health - Health checks - GET /metrics - Prometheus metrics (with .with_monitoring())

Run with: fastapi dev your_file.py → Service ready at http://localhost:8000

Installation

uv add chapkit

Chapkit automatically installs servicekit as a dependency.

Links

• Repository
• Issues
• Servicekit - Core framework foundation (docs)

License

AGPL-3.0-or-later