“Ours is not an industry that respects tradition;
it only respects innovation.” – Satya Nadella
DevOps for AI
Computer on every desk
in every home
To empower every person and every organization
on the planet to achieve more
Build technology platforms and productivity services for an intelligent cloud and intelligent edge infused with AI
AI across Microsoft
Microsoft 365
AI for Good
AutomatedML capabilityin Azure Machine Learning
AI in Office 365
PowerPoint Designer Dictation & read aloud in Office Ideas in Excel
Automated AI in Power BI
Machine Learning on Azure
Domain Specific Pretrained Models
To reduce time to market
Azure
Databricks
Machine
Learning VMs
Popular Frameworks
To build machine learning and deep learning solutions
TensorFlowPytorch Onnx
Azure Machine
Learning
LanguageSpeech
…
SearchVision
Productive Services
To empower data science and development teams
Powerful Infrastructure
To accelerate deep learning
Scikit-Learn
PyCharm Jupyter
Familiar Data Science tools
To simplify model development
Visual Studio Code Command line
CPU GPU FPGA
From the Intelligent Cloud to the Intelligent Edge
AI DevOps lifecycle
Experiment
Data Acquisition
Business Understanding
Initial Modeling
Develop
Modeling
Operate
Continuous Delivery
Data Feedback Loop
System + Model Monitoring
Experiment
+ Testing
Continuous Integration
Continuous Deployment
• Profile and validate
• Data (changes to shape / profile)
• Model in isolation (offline A/B)
• Model + app (functional testing)
• Gate promotion of the model on validation
• Assess
• Functional behavior
• Performance characteristics
• Processes and procedures to make models
reproducible
• Code
• Data
• Config
• Capture
• Featurization code (w/ tests)
• Training pipeline
• Dependencies
• Training data persistence
• Evidence chain
• Model config
• Training job info
• Sample data
• Data profile
• Safe and efficient deployment & feedback
• Simplify consumption - code-generation, API
specifications / interfaces
• Support a variety of inferencing targets
• Cloud Services
• Mobile / Embedded Applications
• Edge Devices
• Convert / quantize / optimize models for
target platform
• Control the rollout of your models (with A/B)
• Feed telemetry back into your system on
service health and model behavior
• Track model versions & metadata with a centralized
model registry
• Leverage containers to capture runtime
dependencies for inference
• Leverage an orchestrator like Kubernetes to provide
scalable inference
• Capture model telemetry – health, performance,
inputs / outputs
• Encapsulate each step in the lifecycle to enable
CI/CD and DevOps
• Automatically optimize models to take advantage of
hardware acceleration
Microsoft DevOps for AI with GoDataDriven
Microsoft DevOps for AI with GoDataDriven

Microsoft DevOps for AI with GoDataDriven

  • 1.
    “Ours is notan industry that respects tradition; it only respects innovation.” – Satya Nadella DevOps for AI
  • 2.
    Computer on everydesk in every home
  • 3.
    To empower everyperson and every organization on the planet to achieve more Build technology platforms and productivity services for an intelligent cloud and intelligent edge infused with AI
  • 4.
  • 7.
  • 8.
    AI in Office365 PowerPoint Designer Dictation & read aloud in Office Ideas in Excel
  • 9.
  • 10.
    Machine Learning onAzure Domain Specific Pretrained Models To reduce time to market Azure Databricks Machine Learning VMs Popular Frameworks To build machine learning and deep learning solutions TensorFlowPytorch Onnx Azure Machine Learning LanguageSpeech … SearchVision Productive Services To empower data science and development teams Powerful Infrastructure To accelerate deep learning Scikit-Learn PyCharm Jupyter Familiar Data Science tools To simplify model development Visual Studio Code Command line CPU GPU FPGA From the Intelligent Cloud to the Intelligent Edge
  • 12.
    AI DevOps lifecycle Experiment DataAcquisition Business Understanding Initial Modeling Develop Modeling Operate Continuous Delivery Data Feedback Loop System + Model Monitoring Experiment + Testing Continuous Integration Continuous Deployment
  • 13.
    • Profile andvalidate • Data (changes to shape / profile) • Model in isolation (offline A/B) • Model + app (functional testing) • Gate promotion of the model on validation • Assess • Functional behavior • Performance characteristics
  • 14.
    • Processes andprocedures to make models reproducible • Code • Data • Config • Capture • Featurization code (w/ tests) • Training pipeline • Dependencies • Training data persistence • Evidence chain • Model config • Training job info • Sample data • Data profile
  • 15.
    • Safe andefficient deployment & feedback • Simplify consumption - code-generation, API specifications / interfaces • Support a variety of inferencing targets • Cloud Services • Mobile / Embedded Applications • Edge Devices • Convert / quantize / optimize models for target platform • Control the rollout of your models (with A/B) • Feed telemetry back into your system on service health and model behavior
  • 16.
    • Track modelversions & metadata with a centralized model registry • Leverage containers to capture runtime dependencies for inference • Leverage an orchestrator like Kubernetes to provide scalable inference • Capture model telemetry – health, performance, inputs / outputs • Encapsulate each step in the lifecycle to enable CI/CD and DevOps • Automatically optimize models to take advantage of hardware acceleration