This paper explores a novel vision for the disciplined, repeatable, and transparent model-driven development and Machine-Learning operations (ML-Ops) of intelligent enterprise applications. The proposed framework treats model abstractions of AI/ML models (named AI/ML Blueprints) as first-class citizens and promotes end-to-end transparency and portability from raw data detection- to model verification, and, policy-driven model management. This framework is grounded on the intelligent Application Architecture (iA 2) and entails a first attempt to incorporate requirements stemming from (more) intelligent enterprise applications into a logically-structured architecture. The logical separation is grounded on the need to enact MLOps and logically separate basic data manipulation requirements (data-processing layer), from more advanced functionality needed to instrument applications with intelligence (data intelligence layer), and continuous deployment, testing and monitoring of intelligent application (knowledge-driven application layer). Finally, the paper sets out exploring a foundational metamodel underpinning blueprint-model-driven MLOps for iA 2 applications, and presents its main findings and open research agenda.
|Number of pages||13|
|Publication status||Published - 2020|
Bibliographical noteDBLP's bibliographic metadata records provided through http://dblp.org/search/publ/api are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.
- AI software engineering
- ML Blueprints
- Methodological support to AI