Skip to main content

Machine Learning Operations Toolkit

Project description

GitHub GitHub GitHub

⏳ Tempo: The MLOps Software Development Kit

Vision

Enable data scientists to see a productionised machine learning model within moments, not months. Easy to work with locally and also in kubernetes, whatever your preferred data science tools

Documentation

Highlights

Tempo provides a unified interface to multiple MLOps projects that enable data scientists to deploy and productionise machine learning systems.

  • Package your trained model artifacts to optimized server runtimes (Tensorflow, PyTorch, Sklearn, XGBoost etc)
  • Package custom business logic to production servers.
  • Build an inference pipeline of models and orchestration steps.
  • Include any custom python components as needed. Examples:
    • Outlier detectors with Alibi-Detect.
    • Explainers with Alibi-Explain.
  • Test Locally - Deploy to Production
    • Run with local unit tests.
    • Deploy locally to Docker to test with Docker runtimes.
    • Deploy to production on Kubernetes
    • Extract declarative Kubernetes yaml to follow GitOps workflows.
  • Supporting a wide range of production runtimes
    • Seldon Core open source
    • KFServing open source
    • Seldon Deploy enterprise
  • Create stateful services. Examples:
    • Multi-Armed Bandits.

Workflow

  1. Develop locally.
  2. Test locally on Docker with production artifacts.
  3. Push artifacts to remote bucket store and launch remotely (on Kubernetes).

overview

Motivating Synopsis

Data scientists can easily test their models and orchestrate them with pipelines.

Below we see two Models (sklearn and xgboost) with a function decorated pipeline to call both.

def get_tempo_artifacts(artifacts_folder: str) -> Tuple[Pipeline, Model, Model]:

    sklearn_model = Model(
        name="test-iris-sklearn",
        platform=ModelFramework.SKLearn,
        local_folder=f"{artifacts_folder}/{SKLearnFolder}",
        uri="s3://tempo/basic/sklearn",
    )

    xgboost_model = Model(
        name="test-iris-xgboost",
        platform=ModelFramework.XGBoost,
        local_folder=f"{artifacts_folder}/{XGBoostFolder}",
        uri="s3://tempo/basic/xgboost",
    )

    @pipeline(
        name="classifier",
        uri="s3://tempo/basic/pipeline",
        local_folder=f"{artifacts_folder}/{PipelineFolder}",
        models=PipelineModels(sklearn=sklearn_model, xgboost=xgboost_model),
    )
    def classifier(payload: np.ndarray) -> Tuple[np.ndarray, str]:
        res1 = classifier.models.sklearn(input=payload)

        if res1[0] == 1:
            return res1, SKLearnTag
        else:
            return classifier.models.xgboost(input=payload), XGBoostTag

    return classifier, sklearn_model, xgboost_model

Save the pipeline code.

from tempo.serve.loader import save
save(classifier)

Deploy to docker.

from tempo.seldon.docker import SeldonDockerRuntime
docker_runtime = SeldonDockerRuntime()
docker_runtime.deploy(classifier)
docker_runtime.wait_ready(classifier)

Make predictions on containerized servers that would be used in production.

classifier.remote(payload=np.array([[1, 2, 3, 4]]))

Deploy to Kubernetes for production.

from tempo.serve.metadata import RuntimeOptions, KubernetesOptions
runtime_options = RuntimeOptions(
        k8s_options=KubernetesOptions(
            namespace="production",
            authSecretName="minio-secret"
        )
    )

from tempo.seldon.k8s import SeldonKubernetesRuntime
k8s_runtime = SeldonKubernetesRuntime(runtime_options)
k8s_runtime.deploy(classifier)
k8s_runtime.wait_ready(classifier)

This is an extract from the introduction demo.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for mlops-tempo, version 0.1.0.dev12
Filename, size File type Python version Upload date Hashes
Filename, size mlops_tempo-0.1.0.dev12-py3-none-any.whl (58.2 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size mlops-tempo-0.1.0.dev12.tar.gz (33.0 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page