DataCanvas APS

One-stop, Low-code, Self-learning

DataCanvas APS (AI Infrastructure Platform Service) provides data scientists, application developers and business experts with a set of tools to process multi-source heterogeneous data autonomously and easily, develop, train and deploy machine learning and deep learning models at any scale quickly and efficiently, to open up the last mile of enterprise-level "large + small" model applications.

观看视频

Product Values Advantages Applications White Paper Success Cases

Product Values

Fully Openess

"White Box" algorithm library
200+Built-in source code
"Trinity" modeling method
Flexible and scalable operating environment

Large Scale Distributed

Multiple distributed training environments
Combined with various training accelerations
Strong support for enterprises LLMs training

Efficient Model Services

Production oriented model services
Integrating lifecycle capability of AI model
Empowering enterprise intelligent applications

Advantages

One stop, high-performance, and highly collaborative

Integrate various internal and external frameworks and tools
Integrate mainstream distributed computing frameworks
Pre-set and automatically build rich distributed training scenarios
Enterprise level multi tenant functionality
Multi personnel collaborative work and result sharing
Heterogeneous multi engine fusion architecture

Flexible computing environment support
Integrate mainstream open-source frameworks, and toolkits
Support the development of multiple languages
Support mixed processing of massive heterogeneous multi-source data
A hybrid workflow orchestration with high flexibility and reusability
ML Engineering & Operation

Built-in model repository, connecting model and production
One click model launch, A/B test/grayscale release
Flexible model services, supporting real-time online and batch services
Comprehensive model services and operational monitoring
Manage third-party models, support model SDK and model file export
AutoML Technology Throughout The Entire Process

Integrate cutting-edge self-developed AutoML open-source software
Visualize model optimization configuration
Automation of all links in the entire model production process
Automatically select，publish，optimize, update the optimal model
Automated modeling function with 0 encoding