Argituml

Automation and explainability of the AI model lifecycle in industrial environments

Description

ARGITUML seeks to promote the effective adoption of artificial intelligence in industrial environments through the deployment of MLOps practices, microservices, and workflows integrated with business solutions. It addresses key challenges such as data scarcity, model drift in production, and user-friendly explainability, with a comprehensive view of the AI model lifecycle.

Objective

To increase the implementation of AI projects in the Basque industrial environment, providing organizations with methodologies and tools to efficiently, explainably, and automatically manage the entire model lifecycle, from training to operation in Edge-Fog-Cloud environments.

Actions

• Application of the MLOps paradigm in industrial environments.
• Generation of synthetic data and mitigation of domain jump.
• Development of Edge-Fog-Cloud architectures for automatic model deployment.
• Detection of drift in predictions and data in production.
• Visual and semantic explainability of time series models adapted to different profiles.
• Implementation of few-shot learning for continuous learning with new data.
• Design of a complete workflow for industrial AI projects.