Speaker: Juan Camilo Díaz
Video: https://youtu.be/jfH93vdRmTk
Kubeflow hace que implementar workflows de Machine Learning en Kubernetes sean simples, portátiles y escalables. Kubeflow es el kit de herramientas que permite implementar procesos de Machine Learning, ampliando la capacidad de Kubernetes para ejecutar pasos independientes y configurables, con bibliotecas y frameworks específicos.
---------------------------------------------------------------------------------------------------------------------------------------------------------------
Hay trabajos y hay carreras. Las oportunidades vienen a golpear la puerta cuando menos lo esperas. La decisión es tuya. Desde tener la oportunidad de hacer algo significativo día tras día, hasta estar rodeado de gente supremamente inteligente y motivada.
¿Estás listo?
Descúbre todas nuestras oportunidades acá: https://bit.ly/2PWKky9
---------------------------------------------------------------------------------------------------------------------------------------------------------------
Síguenos en:
Facebook: https://www.facebook.com/Globant/
Twitter: https://twitter.com/Globant
Instagram: https://www.instagram.com/globantpics/
Linkedin: https://www.linkedin.com/company/globant
2. JUAN CAMILO DÍAZ
4.5 años en Globant
Acerca de mí ...
Juan Camilo Díaz Ortega
Big Data Architect at Globant
Data & Analytics Studio
Kubeflow
3. Why it is so painful to deploy
Machine Learning workflows?
4. Kubeflow Machine Learning Workflow
Gathering data
Data pre-processing
Researching the model that will be
best for the type of data
Training and testing the model
Evaluation
10. Kubeflow Containers
Containers are technologies that allow you to package and isolate
applications along with the entire runtime environment, that is, with
all the files that Containers require to run
Allows you to move the application that is inside the container
between the environments (development, test, production, etc.),
without losing any of its functions.
12. Kubeflow Kubernetes
Kubernetes (also known as k8s or "kube") is an open-source system
for automating deployment, scaling, and management of
containerized applications. Container orchestration platform.
In other words, you can cluster together groups of containers, and
Kubernetes helps you easily and efficiently manage those clusters.
Kubernetes clusters can span hosts across on-premise, public,
private, or hybrid clouds. For this reason, Kubernetes is an ideal
platform for hosting cloud-native applications that require rapid
scaling
13. Kubeflow Kubernetes
Orchestrate containers across multiple
hosts.
Scale containerized applications and
their resources on the fly
Control and automate application
deployments and updates
Health-check and self-heal your apps
with autoplacement, autorestart,
autoreplication, and autoscaling.
14.
15. DefinitionKubeflow
Kubeflow is an open source Kubernetes-native platform for developing,
orchestrating, deploying, and running scalable and portable machine learning
workloads
Portable Machine Learning Stack
The Kubeflow project is dedicated to making deployments of machine learning
(ML) workflows on Kubernetes simple, portable and scalable.
https://www.kubeflow.org/docs/about/kubeflow/
30. PipelinesKubeflow
Kubeflow Pipelines is a platform for
building and deploying portable and
scalable end-to-end ML workflows,
based on containers.
Code that performs one step in the
Pipeline. In other words a
containerized implementation of an
ML task.
31. PipelinesKubeflow
A pipeline is a description of an
ML Workflow
It runs a containers which
provide portability, repeatability
and encapsulation, which is able
to decouples the execution
environment to code runtime.
40. Katib - Hyperparameter TuningKubeflow
Hyperparameters are the variables that
control the model training process. For
example:
● Learning rate.
● Number of layers in a neural
network.
● Number of nodes in each layer.
46. Cloud Computing - Cloud Providers
12 months of popular free services
+
$200 credit to explore Azure for 30 days
+
Always free 25+ services
https://azure.microsoft.com/en-us/free/
12 months free services
+
Short-term free trial offers start from the date
you activate a particular service
+
Always free, free tier offers do not expire and
are available to all AWS customers
https://aws.amazon.com/free/
12 months free services
+
$300 free credit
+
Always free products, which provides limited
access to many common Google Cloud
resources, free of charge.
https://cloud.google.com/free
https://docs.microsoft.com/en-us/azure/ https://cloud.google.com/docs https://docs.aws.amazon.com/index.html
Kubeflow