Google Cloud AI Platform Introduction

AI Platform provides managed services you need to build and deploy a model from scratch

-Source and prepare your data
Uses Pub/Sub, Dataflow, and Dataprep to ingest, prepare and transform data

-Code, train and evaluate your model
Uses AI Platform Notebooks and Training services to build, and evaluate models
Coding use pre-trained models and/or TF modules, notebooks from AI hub

-Deploy and get predictions from your trained model
Uses AI Platform Predictions to serve models, and Kubeflow Pipelines to encapsulate ML workflows for reuse

Transcript

1. Bring the World Smarter Together since 2017 Taiwan ‧ Hong Kong ‧ Singapore Google Cloud AI Platform Introduction
2. Copyright © 2019, CloudMile Who I am —— Brent Chang @ CloudMile •GCP/AWS Certified Cloud Architect •4yr+ Public Cloud experience •“Stay Hungry, Stay foolish!”
3. Copyright © 2019, CloudMile 1st GCP Infra & ML Specializations N.Asia. 40+ Certifications Top10 TW Coolest Startup We enable enterprises to turn AI into real opportunity. 200+ Cloud Customers in HK, TW
4. Copyright © 2019, CloudMile Confidential Copyright © 2019, CloudMile Custom-code prediction Disease fast screening Visual search Content recommender system Recommender system for e-commerce Anomaly detection Defects inspection Logistic Healthcare Textile Media Manufacturing Gaming e-commerce FSI Intelligent document recognition
5. Welcome to join us! https://www.facebook.com/groups/GCPUG.TW/ https://www.meetup.com/GCPUG-Taipei/
6. www.companyname.com © 2016 Startup theme. All Rights Reserved. PAGE www.mile.cloud 6 Copyright © 2019, CloudMile Google Cloud AI Platform Introduction
7. Copyright © 2019, CloudMile Machine Learning nowadays • Preparing the data is time-consuming task • Hard to deploy your model to production • Hard to reuse or share with other users
8. Copyright © 2019, CloudMile Machine Learning nowadays • Preparing the data is time-consuming task • Hard to deploy your model to production • Hard to reuse or share with other users
9. Copyright © 2019, CloudMile Machine Learning workflow
10. Copyright © 2019, CloudMile Source and prepare your data Ingest Transform Cloud Dataflow Ingest and distribute data reliably Fast, correct computations quickly and simply Cloud Pub/Sub Label AI Platform Data Labeling High quality training data for machine learning
11. Copyright © 2019, CloudMile Source and prepare your data
12. Copyright © 2019, CloudMile Source and prepare your data
13. Copyright © 2019, CloudMile Code your model Configured environments Built-in algorithms AI Platform Get a head start with fully configured VMs for popular data science and deep learning frameworks Use the built-in algorithms inside AI Platform to develop with and deploy Deep Learning VM ML Pipelines and more AI Hub Build onto end-to-end ML pipelines, notebooks, and related content from the AI Hub
14. Copyright © 2019, CloudMile Deep Learning VM image Fast prototyping Prototype your ML project quickly with pre-configured VMs for deep learning. CPU, GPU and TPU support Choose to add the latest Cloud TPU or GPUs on Google Cloud to your instance in a single click and accelerate your model training jobs. Performance optimized for Google Cloud We tune the libraries and config to get the optimal performance on our infrastructure, so you don’t need to worry about it. Flexibility Choose between different ML frameworks like TensorFlow, PyTorch, and scikit-learn or install your own on top of our common base image.
15. Copyright © 2019, CloudMile Deep Learning VM image —— One-Click Deployment Get started quickly Spin up a JuypterLab instance, pre-configured with the latest Machine Learning and data science frameworks in one click. AI Platform
16. Copyright © 2019, CloudMile Serverless training using AI Platform • Train models without managing infrastructure • Supports all popular data science and machine learning frameworks. • Leverage distributed training on the latest GPUs and TPUs to finish jobs faster • Improve your model quality with the state-of-the-art automated hyperparameter tuning
17. Copyright © 2019, CloudMile Easy deployment on GCP and hybrid Deploy pipelines via Kubeflow on GCP and on premise. AI Hub —— One stop AI catalog AI HubBETA One stop AI catalog Easily discover plug & play pipelines & other content built by Google AI and partners. Private hosting Host pipelines and ML content with private sharing controls within an enterprise to foster reuse within organizations..
18. Copyright © 2019, CloudMile AI Hub —— One stop AI catalog AI Hub Public Content + Private Content By Google Unique AI assets by Google By Partners Created, shared & monetized by anyone. By Customers Content shared Securely within and with other organizations. AutoML, TPUs, Cloud AI Platform, etc.
19. Copyright © 2019, CloudMile On-premises training using Kubeflow Infrastructure abstraction Kubernetes manages all underlying dependencies, and resources Swappable & scalable Library of ML microservices to deploy training and prediction jobs Run where you want • GCP • On-premises ML microservices Cloud On- premises Training Predict Training Predict … … Kubernetes
20. Copyright © 2019, CloudMile Deploy your model with ease • Set up online endpoints for low-latency predictions, or get predictions on massive batches of data • Deploy models trained on premises or on Google Cloud • Scale automatically based on your traffic • Use GPUs for faster predictions
21. Copyright © 2019, CloudMile What is included in AI Platform? AI Platform Integrated with Deep Learning VM Images Cloud Dataflow Cloud Dataproc Google BigQuery Cloud Dataprep Google Data Studio Notebooks Data Labeling Training Predictions Built-in Algorithms For data warehousing For data transformation For data cleansing For Hadoop and Spark clusters For BI dashboards Kubeflow (On premises) AI Hub
22. Copyright © 2019, CloudMile How about on-prem? Connect to database Create datasets & labels Feature engineering Train scikit-learn linear regression model HP tuning Publish model Business logic Connect to database Create datasets & labels scikit-learn linear regression Pipeline Business logic Data Engineer + Developer Today Reusable Kubeflow Pipelines Data Engineer + Developer Data Scientist + ML Engineer Developer
23. Copyright © 2019, CloudMile Which supports the entire Data Science team Data Engineer Uses Pub/Sub, Dataflow, and Dataprep to ingest, prepare and transform data Data Scientist Uses AI Platform Notebooks and Training services to build, and evaluate models ML Engineer Uses AI Platform Predictions to serve models, and Kubeflow Pipelines to encapsulate ML workflows for reuse. Developer Collaborates with data scientists to embed AI through REST APIs into applications Business Analyst Discovers solutions from AI Hub and deploys it into production
24. Copyright © 2019, CloudMile Get Started with Google Cloud AI Platform Discover pipelines and notebooks from AI Hub cloud.google.com/ai-hub Start building with our fully-configured, managed notebooks cloud.google.com/ai-platform-notebooks Try tutorials, quickstarts, and more. cloud.google.com/ai-platform
25. Thank You! CloudMile Facebook brent.chang@mile.cloud