Build & Deploy AI Models End-to-End with PAI DSW and EAS - Platform for AI

Prerequisites

Before activating PAI, verify the following requirements:

Account permissions: Use an Alibaba Cloud account or RAM user with AliyunPAIFullAccess policy attached.
Account balance: Ensure account balance is positive (≥0).

Region availability: PAI is available in the following regions:

Region	Region ID
China (Beijing)	cn-beijing
China (Hangzhou)	cn-hangzhou
China (Shanghai)	cn-shanghai
China (Shenzhen)	cn-shenzhen
Singapore	ap-southeast-1
US (Silicon Valley)	us-west-1
Germany (Frankfurt)	eu-central-1

Activate PAI

Log on to the PAI console.
Select a region from the dropdown list.
Click Activate Now.

A default workspace is created automatically upon activation.

Note: RAM users without AliyunPAIFullAccess permissions will encounter activation failures. Attach the policy before retrying.

Explore functional modules

Select a module based on your development needs.
Follow the quick start guide to learn basics through hands-on examples.
Explore the user guide for advanced features and best practices.

Note: Click the module name to access the complete user guide.

Module	Description	Quick start
Data Science Workshop (DSW)	Cloud-based IDE for AI development. Developers familiar with Jupyter Notebook or VSCode can start model development immediately.	DSW Quick Start
Machine Learning Designer	140+ algorithm components for low-code visual modeling through drag-and-drop interface.	Designer Quick Start
Deep Learning Containers (DLC)	Runs distributed or standalone training tasks without manual machine provisioning or environment configuration.	DLC Quick Start
Elastic Algorithm Service (EAS)	Deploys trained models as online inference services.	EAS Quick Start
Model Gallery	Trains and deploys open-source large language models with zero code.	Model Gallery Quick Start

Troubleshooting

Error 100900010: Service activation failed

Symptom: Activation fails with error code 100900010, particularly in Germany (Frankfurt) region.

Possible causes:

RAM user lacks required permissions.
Account balance is negative.
Region quota exceeded or service temporarily unavailable.

Solutions:

Verify RAM user has AliyunPAIFullAccess policy attached. See Grant permissions to a RAM user.
Check account balance at User Center. Pay outstanding fees if balance is negative.
Try activating in a different region from the supported regions list.
If error persists, submit a ticket with the following information:
- Alibaba Cloud account ID
- Region where activation failed
- Timestamp of activation attempt
- Complete error message and code

Activate Now button is grayed out

Symptom: Button shows message "The current account does not have permission to activate PAI. Contact the owner of your Alibaba Cloud account."

Solutions:

Use your Alibaba Cloud account to complete activation.
For RAM users, attach the AliyunPAIFullAccess system policy. Note: This policy grants extensive permissions. The account administrator should assess security risks before granting.

Account balance error during activation

Symptom: Error message Create order error: message is Your account balance is less than 0. Please top up your account and try to purchase again. productRequestId is *** appears.

Solution: Visit the User Center to view bills and pay outstanding fees, then retry activation.

Get help

FAQ documents contain common questions and solutions from developers. If you encounter issues with modules like DSW or EAS, refer to the corresponding FAQ document:

DSW FAQ	EAS FAQ	DLC FAQ	Feature Store FAQ
Model Training FAQ	Designer FAQ	Dataset FAQ	Billing FAQ

Typical AI development workflows

PAI covers the complete AI development lifecycle from data preparation and model training to deployment. The following sections describe two typical workflows.

Cloud-native AI development

Step	Description	References
①	Dataset management centrally manages local, cloud, and public datasets as data sources for model training.	Dataset management overview
②	DSW provides a cloud-based IDE for AI development. Developers familiar with Jupyter Notebook or VSCode can start model development immediately.	DSW overview
③	Image management centrally manages official images and custom images, providing runtime environments for your code.	Image management
④	After developing and testing model code in DSW, use DLC to run training tasks.	Create a DLC training task
⑤	PAI supports mounting file systems (NAS and OSS) and Git repositories to simplify data and code specification when submitting tasks.	Code management
⑥	Model management centrally manages trained models for deployment using EAS.	Model management
⑦	Deploy the trained model as an online inference service using EAS.	Deploy an EAS model service

AI and big data development

Step	Description	References
①	For data stored in MaxCompute, use DataWorks for data preprocessing, then reference the MaxCompute table as training data source in PAI.	General data development
②	Designer offers 140+ algorithm components for low-code visual modeling through drag-and-drop interface.	Visual Modeling Designer
③	Configure and run scheduled tasks using DataWorks.	Schedule PAI nodes
④	Task management records execution details for experiments and custom tasks, simplifying task comparison and analysis.	Task management
⑤	Model management centrally manages trained models for deployment using EAS.	Manage models
⑥	Deploy the trained model as an online inference service using EAS.	Deploy an EAS model service

Next steps

Develop models interactively: DSW Quick Start
Train models at scale: DLC Quick Start
Deploy inference services: EAS Quick Start