MetaSpore one-stop machine learning framework
An intelligent machine learning framework that breaks through the gap between big data framework and AI platform, AI platform and actual business landing
One-stop design concept and solutions
It leverages standardized components and development interfaces to provide a one-stop development experience that meets the needs of enterprises and developers to capture algorithmic business development best practices.
Model training integrated with big data systems seamlessly
It can directly read the structured and unstructured data of various data lakes and silos for training and seamlessly connect the data, feature preprocessing, and model training, avoiding the tedious process of data import, export, and format transformation.
Support for sparse features
It applies to training scenarios such as the large-scale sparse Embedding layer, cross-combination, and variable length feature pooling.
Provides high-performance online inference service
Neural networks (including sparse Embedding), decision trees, and various traditional machine learning models are supported. It supports heterogeneous hardware computing acceleration and reduces engineering barriers for online deployment.
Unified offline feature computation
Through the unified feature format and calculation logic, the suitable offline feature calculation saves the cost of repeated development of multiple systems and ensures the consistency of offline features.
Online algorithm application framework
It covers the common function points of online systems, such as automatic feature extraction from multiple data sources, feature computation, predictive service interface, dynamic configuration of experiments, and dynamic stream cutting of ABTest.
Embrace open source
It provides multiple self-developed components to realize the core functions and embraces the mature open ecology to reduce the learning threshold.
One-stop multimodal AI platform
Break through the "end to end" of the process, one-stop solution to a variety of business problems
Business Application Framework
Paving last mile of algorithm models' application to business. Provides AB testing, visualized debugging and parameters hot reload.
Online Inference Framework
A high performance, low latency model inference service for large scale models, which supports heterogeneous hardwares and can be deployed as a micro service.
Offline Training Framework
Fill the gap between data and AI. Supports streaming online learning with higher efficiency and accuracy.
Solution: Low-code Personalized Recommendations
Bert, Swing, Item CF, and other machine learning algorithms are used to optimize based on different industries and massive data, with high accuracy and pertinence. Non-ai professionals can deploy quickly with low code and improve key business metrics such as click-through and conversion rates.
You can import all data quickly and efficiently with a few operations. The industry-grade recommendation system brings users different homepage experiences -- guess what you like, look and look, post-purchase link, and other scenarios, enabling users to discover products faster and more easily, creating first-class browsing and consumption experience, promoting purchase decisions, and achieving business growth.
Join the community and share data intelligence