Thematic categorization using LLM to build action plans (F/H) - intership

The objective of this internship will be to develop technological bricks associated to thematic categorization. Some bricks will involve Generative AI and LLMs, with the help of AI teams. Other bricks will involve automated pipeline construction, with the help of MLOPs teams. During the projects interactions between end users with the Business Units and Functions involved will be done to make sure the solution developed is adapted to a real need.

The results of this internship will be presented to Saint-Gobain AI teams and Saint-Gobain Business Units involved in the project.

Core Responsabilities

• - Extraction of main thematics in a set of multilingual verbatims using LLM


• – Thematic classification of multilingual verbatims using LLMs and ADA embeddings on different use-cases


• – Analytics and visualizations on the thematic extracted for executive summaries


• – Building automated pipelines (with the help of MLOPs experts) to trigger thematic categorization on specific events (e.g. when a Microsoft Forms is submitted)


• - Building simple interfaces using Streamlit to show results to end-users.

Technologies we use

• Python


• Azure Machine Learning 

Generative AI - Open AI secured by Azure -  Chat GPT / ADA embedding