Development and industrialization of AI-based solutions under MLOPs/LLLMOPs principles

Outline

In this seminar you will learn about the different modalities of application and industrialization of Artificial Intelligence based solutions to support the internal processes of an organization through the strategic, functional and analytical description of two use cases:

• A solution that helps university professors to speed up attention by generating draft responses to student emails, considering context information both from the student themselves and from their past interactions with the university. This solution is cloud-based and does not require fine-tuning.
• Application to support the classification of protected documents through the discriminative use of LLMs, which allows to categorize documents in a taxonomy of 70 categories and subcategories, through a solution based on Open Source algorithms with fine-tuning, developed in an on-premise architecture, due to the privacy of the processed documents.

In this space you will get an introduction to MLOPs and LLMOPs to set the principles that must govern the industrialization of an analytical solution. Then, how these principles were applied in the construction of an industrialized environment will be detailed.

Assessment Method

Participation in the class plus an individual assignment.

Lective hours

Timetable

• 21 March, 15:00-18:00

Lecture Theatre

• A-6202

Tuition Language

Spanish.