In industries represented by financial institutions, clients have fully transitioned from informatization and digitalization to intelligence. Intelligence has become an essential part of enterprise transformation and upgrading. AI empowers enterprise decision-making, process automation, and convenient interaction, becoming the driving force for future corporate competition. Financial institutions need to layout a comprehensive intelligent upgrade from a global enterprise perspective, supporting business strategies. Therefore, they require a unified intelligent platform to produce and manage various AI algorithms within the organization, providing functionalities ranging from data processing, algorithm development, model training, model deployment, and model management, serving as the engine for enterprise intelligence.

The ORIGIEN Industry AI Platform, addressing the high barriers, long cycles, and difficulties of implementing AI technology development, uses cutting-edge technology to create a full-link, end-to-end AI algorithm development suite and tools, enabling low-barrier development of complex models. At the same time, it builds an enterprise-level AI service middle platform, standardizing and unifying AI model services, fulfilling industry-wide needs such as "convenient algorithm development, ultra-large-scale AI training, AIPaaS services, and AI security and trustworthiness," helping enterprises achieve intelligent upgrades.

1、AI Service Platform

Supports real-time inference API services and batch prediction processing services for traditional machine learning models and deep learning models. It also supports online governance and orchestration of models, including model monitoring, early warning, rolling updates, as well as features such as load balancing, elastic scaling, and circuit breaking. Additionally, it supports model service orchestration, workflow management, and AI application-level service orchestration.

2、AI Development Platform

Provides a full-link AI algorithm development suite for algorithm developers, suitable for different user modeling needs. It enables one-stop, convenient algorithm development, covering data labeling, feature engineering, model development (AutoML, visual drag-and-drop modeling, notebook programming modeling), model evaluation, and model deployment. It also includes rich algorithms and industry algorithm experience templates, supports visualization of model training processes, unified model management, model asset management, and supports model deployment and grayscale publishing.

3、AI Computing Framework

Facilitates model computation and training. It supports mainstream deep learning frameworks such as TensorFlow, PyTorch, Keras, MxNet, and is compatible with mainstream big data computing frameworks like Spark and Flink. It also supports mixed parallel modes such as data parallelism, model parallelism, and pipeline parallelism, as well as distributed training for ultra-large models under the large model trend.

4、AI Computational Power Platform

Virtualizes underlying AI computing resources and manages and schedules AI computing resources centrally to improve resource utilization. Its features include GPU resource virtualization, heterogeneous AI computing resource management, task scheduling, and service scheduling.

5、AI Security Center

Ensures the safety of AI model production and operation through robustness testing, adversarial training, model poisoning monitoring, and adversarial sample monitoring, ensuring that intelligent business applications are secure and trustworthy.