Download as PDF, PPTX
Uploaded byMárton Kodok
Vertex AI: Pipelines for your MLOps workflows
The document discusses Vertex AI pipelines for MLOps workflows. It begins with an introduction of the speaker and their background. It then discusses what MLOps is, defining three levels of automation maturity. Vertex AI is introduced as Google Cloud's managed ML platform. Pipelines are described as orchestrating the entire ML workflow through components. Custom components and conditionals allow flexibility. Pipelines improve reproducibility and sharing. Changes can trigger pipelines through services like Cloud Build, Eventarc, and Cloud Scheduler to continuously adapt models to new data.
More Related Content
Similar to Vertex AI: Pipelines for your MLOps workflows
Vertex AI: Pipelines for your MLOps workflows
- 1. Vertex AI Pipelines foryour MLOps workflows GDG DevFest, November 2021 Márton Kodok Google Developer Expert at REEA.net
- 2. ● Among theTop3 romanians on Stackoverflow 195k reputation ● Google Developer Expert on Cloud technologies ● Crafting Web/Mobile backends at REEA.net ● BigQuery + Redis database engine expert Slideshare: martonkodok Articles: martonkodok.medium.com Twitter: @martonkodok StackOverflow: pentium10 GitHub: pentium10 Vertex AI: Pipelines for your MLOps workflows @martonkodok About me
- 3. 1. What isMLOps? 2. What is Vertex AI? 3. Build, train and deploy ML solutions 4. Using Pipelines throughout your ML workflow 5. Adapting to changes of data 6. Conclusions Agenda Vertex AI: Pipelines for your MLOps workflows @martonkodok
- 4.
- 5. “ DevOpsprinciples toMLsystems Vertex AI: Pipelines for your MLOps workflows @martonkodok What is MLOps?
- 6. Elements for MLsystems Adapted from Hidden Technical Debt in Machine Learning Systems. @martonkodok
- 7. “Continuousdelivery and automationpipelines formachinelearning systems. Vertex AI: Pipelines for your MLOps workflows @martonkodok What is MLOps?
- 8. MLOps level 0:Manual process MLOps level 1: ML pipeline automation MLOps level 2: CI/CD pipeline automation Levelsofautomation defines maturity of theMLprocess @martonkodok
- 9. MLOps level 0:Manual process - Process for building and deploying ML models is entirely manual. Infrequent release iterations. No CI, No CD. Disconnection between ML and operations. MLOps level 1: ML pipeline automation - Continuous training of the model by automating the ML pipeline; achieve continuous delivery of model prediction service. New pipelines mostly based on new data. MLOps level 2: CI/CD pipeline automation -iteratively try out new ML algorithms and new modeling where the experiment steps are orchestrated. The output of this stage is the source code of the ML pipeline steps that are then pushed to a source repository. Build source. Run test. Output is pipeline. Levelsofautomation defines maturity of theMLprocess @martonkodok
- 10. MLOps level 2:CI/CDpipelineautomation @martonkodok
- 11. Levelsofautomation defines maturityof theMLprocess
- 12.
- 13. “VertexAI is amanaged ML platform for practitioners to accelerate experiments and deploy AI models. Vertex AI: Pipelines for your MLOps workflows @martonkodok
- 14. What’s included inVertexAI? Vertex AI: Pipelines for your MLOps workflows @martonkodok Data Labeling AutoML DL Environment (DL VM + DL Container) Prediction Feature Store Training Experiments Data Readiness Feature Engineering Training/ HP-Tuning Model Monitoring Model serving Understanding/ Tuning Edge Model Management Notebooks Pipelines (Orchestration) Explainable AI Hybrid AI Model Monitoring Metadata Vision Translation Tables Language Video AI Accelerators Models Datasets Custom Models Containers Python Endpoints Vizier Optimization
- 15. VertexAI is aunified MLOps platform Vertex AI: Pipelines for your MLOps workflows @martonkodok Operational Model Programming Model No Infra Management Managed Security Pay only for usage Model-as-a-service oriented Streamlined model development Open SDKs, integrates with ML frameworks
- 16. Using Pipelines throughout your MLworkflow Part #3
- 17. VertexAI: Pipelines -Orchestrate your ML workflow Vertex AI: Pipelines for your MLOps workflows @martonkodok Data Labeling AutoML DL Environment (DL VM + DL Container) Prediction Feature Store Training Experiments Data Readiness Feature Engineering Training/ HP-Tuning Model Monitoring Model serving Understanding/ Tuning Edge Model Management Notebooks Pipelines (Orchestration) Explainable AI Hybrid AI Model Monitoring Metadata Vision Translation Tables Language Video AI Accelerators Models Datasets Custom Models Containers Python Endpoints Vizier Optimization
- 18. “ Why areMLpipelines useful? Vertex AI: Pipelines for your MLOps workflows @martonkodok
- 19. 1. Orchestrate MLworkflow steps as a process. We no longer handle all data gathering, model training, tuning, evaluation, deployment as a monolith. 2. Adopt MLOps for production models. We need a repeatable, verifiable, and automatic process for making any change to a production model. 3. Develop steps independently -as you scale out, enables you to share your ML workflow with others on your team, so they can run it, and contribute code. Enablesyoutotracktheinputandoutputfromeach stepinareproducibleway. Why are ML pipelines useful? @martonkodok
- 20. Vertex AI: Pipelines VertexAI: Pipelines for your MLOps workflows Source: Piero Esposito https://github.com/piEsposito/vertex-ai-tutorials
- 21. Using Pipelines throughoutyour ML workflow Vertex AI: Pipelines for your MLOps workflows @martonkodok Gather data Train model Deploy model
- 22. Pipeline Components Vertex AI:Pipelines for your MLOps workflows pipeline_components_automl_images.ipynb github.com/GoogleCloudPlatform/vertex-ai-samples
- 23. Using Pipelines throughoutyour ML workflow Vertex AI: Pipelines for your MLOps workflows @martonkodok Gather data Train model Evaluate model Scalably deploy model
- 24. Pipeline SDK: Condition VertexAI: Pipelines for your MLOps workflows automl_tabular_classification_beans.ipynb github.com/GoogleCloudPlatform/vertex-ai-samples
- 25. 1. Use ofthe Google Cloud Pipeline Components, which support easy access to Vertex AI services 2. Custom Components - function that compiles to a task ‘factory’ function that can be used by pipelines 3. No more Kubeflow Pipelines that must be deployed on a Kubernetes Cluster. 4. Sharing component specifications - the YAML format allows the component to be put under version control and shared with others, or be used by other pipelines by calling the load_from_url function. 5. Leveraging Pipeline step caching to develop and debug 6. Vertex AI Metadata service + Artifacts Lineage tracking - inverse of pipeline DAG Developer friendly components @martonkodok
- 26. Recap Vertex AI: Pipelinesfor your MLOps workflows @martonkodok Data Labeling AutoML DL Environment (DL VM + DL Container) Prediction Feature Store Training Experiments Data Readiness Feature Engineering Training/ HP-Tuning Model Monitoring Model serving Understanding/ Tuning Edge Model Management Notebooks Pipelines (Orchestration) Explainable AI Hybrid AI Model Monitoring Metadata Vision Translation Tables Language Video AI Accelerators Models Datasets Custom Models Containers Python Endpoints Vizier Optimization
- 27. Part #4 Adapting to changesof data
- 28. Automatic CI /CD Perspective with GCP Services Vertex AI: Pipelines for your MLOps workflows @martonkodok Eventarc • Detect changes on data • React to events from Cloud services • Handle events on Cloud Workflows, Cloud Functions, Cloud Run • Reuse pipeline spec.json from GCS • Trigger Vertex AI pipeline • Detect changes in codebase • Build pipeline • Pipeline spec.json to Cloud Storage • Image to Cloud Registry • Trigger Vertex AI pipeline Cloud Build Cloud Scheduler • Poll for changes of any data • Launch based on schedule • In tandem with Cloud Workflows • Trigger Vertex AI pipeline
- 29. Conclusion Vertex AI: Pipelinesfor your MLOps workflows @martonkodok
- 30. 1. Build withthe groundbreaking ML tools that power Google 2. Approachable from the non-ML developer perspective (AutoML, managed models, training) 3. Ease the life of a data scientist/ML (has feature store, managed datasets, endpoints, notebooks) 4. End-to-end integration for data and AI with build pipelines that outperform and solve complex ML tasks 5. GitOps-style continuous delivery with Cloud Build 6. Explainable AI and TensorBoard to visualize and track ML experiments Vertex AI: Enhanced developer experience Vertex AI: Pipelines for your MLOps workflows @martonkodok
- 31. Thank you. Q&A. Slidesavailable on: slideshare.net/martonkodok Reea.net - Integrated web solutions driven by creativity to deliver projects.