Kubeflow
Google developed Kubeflow, a machine learning stack for its popular TensorFlow ML frame-
work. It is designed to simplify and scale the framework-agnostic modeling, training, serving
and management of containerized AI models across Kubernetes multicloud based ecosystems.
AI-driven intelligence can be thoroughly embedded in every edge, hub and cloud service.
This has made it easier to set up and productionize ML workloads on Kubernetes. It changes
the game by allowing engineers to consistently deploy the entire life cycle of a model, starting
from setting up Jupyter Notebooks and training environments, to packaging and serving the
trained models on production environments using a single framework. Kubeflow abstracts the
underlying resources, and the same deployment works on any environment.
Hyper
parameter
tuning
Validated
Model
Training
model
Tensorflow
PyTorch
Serving
Frameworks
MxNet
Ingress
Controllers
Training Frameworks
Data
Scientists
Model Developmental tools
API
Microservices
DevOps
CUDA drivers
CUDA drivers
NVIDIA GPU
NVIDIA GPU
Model Training
Model Serving
Datasources
Training
Data
Real time
Data
Amazon S3
Azure Blog
Storage
RDS
IOT Devices
Figure 1: Machine Learning Using Kubeflow
50 | THE DOPPLER |
Webapp
Microservice
WINTER 2019
Users