IBM Cloud Platform Overview for Data Scientists
Thomas Schaeck Distinguished Engineer, Watson Studio Data Science and ML in Enterprises Data must be cleaned and shaped for training
need good data avoid bias Data … then models must be created
Enterprises generate TONS OF DATA … requiring cataloging and governance SPSS
…that must be hosted and monitored at scale … and trained on high performance compute
to select an optimal model... IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 2 Why are enterprises struggling to capture the value of AI?
Data Access Data Governance Skills Tools&Infrastructure
• Data resides in silos • If the data isn’t • Data Science skills • Need an environment & difficult to access secure, self-service are in low supply and that enables quick isn’t a reality high demand experiments and real • Unstructured and outcomes external data wasn’t • Challenge • Nurturing new data considered understanding data professionals is • Discrete tools lineage and getting to challenging present barriers to 3 IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation a system of truth productivity Watson AI and Data
Watson Applications, solutions Healthcare Financial Logistics Virtual agent and services Services Studio & IoT Targeted solutions for Knowledge enterprise businesses Catalog
Compare and Watson Machine Tone Personality Assistant Discovery API API API API Watson AI Comply Learning API Analyzer API Insight Cognitive building blocks for developers Nat Language Nat Language Visual Recognition Speech Document …more API API API Conversion API Understanding API Classifier
Data Tools to prepare data Ingestion Storage Analytics Deployment Governance for AI and data persistence
Cloud Integration Cognitive Micro-services DevOps Tooling Cloud Infrastructure A highly scalable, security Networking Compute Security Dedicated Compute Virtual Compute Object Storage …more enabled infrastructure
Public Hybrid Private Collecting, organizing, analyzing Data à Training and deploying Models à Managing AI Ops
Process/Business Owner Data Scientists/Engineers, Manage and oversee IBM + Business Partners Domain Experts models in production provide assets and services analyze data, gain insights, train models AI OpenScale Watson Studio Desktop Watson Studio Gallery Cloud/Local available soon and Community Cloud Manage AI Operations Create personal Projects Assets to get started with - Fairness Exchange Assets with teams through • Notebooks. Models, Dashboards, … - Explainability Cloud/Local Projects and/or Git Repos • Soon: Project Templates - Evolution Deploy Models to WML • - Health and Accuracy Community to communicate and connect End Users Create project benefit Contribute and collaborate from AI API API Assets Watson ML Reuse Assets Apps Watson Studio Projects Watson Knowledge Catalog Use Cloud/Local from Community Processes Models Cloud/Local Cloud/Local Devices Deploy, manage, retrain Models Deploy Web Sites Work with Data and AI Reuse Assets Share Enterprise Assets Deploy Notebooks, Scripts Feed- • Process/cleanse Data from Catalog • Data ... Health and Accuracy back • Create/run Notebooks • Connections • Analyze/visualize Data • Notebooks Use models • Create Dashboards • where they Models Customer’s Repo • Train AI/ML/DL Models • Dashboards are needed • … Export • Soon: Project Templates and Build Process Publish Assets Model as a team or individual. • to Catalog … Consume models e.g feed Intuitive collaboration governed at scale into custom build process Optionally integrate with Git to deploy to production Embed Models Models Models Add/connect Data
Process Designer or App Developer Business Owner Data Owner or Receive insights Data Engineer integrates models into apps or processes IBM Other On as Dashboards, makes data available Cloud Clouds Prem Notebooks, …
Connect to Data in IBM Cloud Data Services, On Prem Data, or Data on other Clouds
Store your data in the IBM Cloud and/or connect your on-prem data or third-party cloud data
...... IBMIBM Cloud Cloud SQL Query Query – Very – HighServerless Level Architecture SQL aaS (MVP 1Q 2018)
Application • Completely self-service SQL on cloud data • Serverless scale-out SQL execution on TB of data 4. Read 1. Submit SQL results • Pay per query (pennies per query) • Full ANSI SQL support, incl. OLAP functions & Spatial SQL Query • Easy SQL operationalization for developers SQL • REST API, Cloud Function integration, SDKs (Python, Node.js)
2. Read 3. Write data results
IBM Cloud Object Storage
Archive & Export Land
IBM Cloud Databases IBM Cloud Streaming Db2 on Cloud Watson IoT
IBM Streams
Streaming Events Eliminates the burden Allows you to focus on IBM Cloud Databases of managing your business A fully managed collection of open infrastructure and objectives rather than system health as well the operations and source databases and persistence as the associated support of your own services available through a consistent capital expenses platform stack consumption, pricing, and interaction model
Avoid vendor lock-in and allows for easily coordinating multi-database strategies by using Open Source DBaaS like:
• Postgres • etcd
• Mongo • MySQL
• ElasticSearch • ScyllaDB
• Redis • JanusGraph
IBM Cloud / V2 / 5-18-2018 / © 2018 IBM Corporation • RabbitMQ 9 Where to get started ? New: Watson Studio Sign Up Page for Germany / EU: https://eu-de.dataplatform.cloud.ibm.com
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 11 Sign Up or use existing IBM Cloud Account – Watson Studio + Knowledge Catalog + Cloud Object Storage will be created for you
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 12 Create your own Projects to connect, analyze, visualize Data and create and train Models
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 13 Watson Studio and Watson Machine Learning in the IBM Cloud Catalog
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 14 Object Storage
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 15 Database Services
IBM Cloud / Watson and Cloud Platform / © 2018 IBM Corporation 16