Design and evaluation of an enterprise Generative AI chat and various systems engineering activities

The SCK CEN IST Institute, in the framework of the MYRRHA project and the ongoing Small Modular Reactor (SMR-LFR) technology demonstrator, has been accruing a considerable amount of knowledge that is being collected into significantly large knowledge bases (KBs) of documents as well as requirements, design decisions, and constraints hosted in Polarion spaces. Due to the increasing size of said KBs, SCK CEN is considering to create an “intelligent chat” able to interact with querying users and answer complex questions about vast and growing sets of information with approaches based on Large Language Models (LLMs) and Retrieval Augmented Generation (RAG). A first prototypic implementation has been designed and prototyped in Python capitalizing on modern modules such as LangChain and LangGraph, though the design is still incomplete and no quality-vs-costs assessment has been carried out. The present internship primarily focuses on the above sketched activity.

Also in the framework of our SMR-LFR activities, a tool is currently being designed at SCK CEN to manage the key performance parameters at the core of design requirements. The tool is available as a prototypic Python + Javascript application that manages parameters sheet offering simple mechanism of change notification when a local change implies repercussions in other local loci and at global level. A second activity of this internship focuses on the completion, fine-tuning, and testing of said application.

Thirdly, sporadic activities shall address the updating of our KBs.

The ideal candidate for this position should be familiar with programming in Python in a distributed environments well as with Python modules for the creation of Generative AI services. Basic knowledge of Javascript shall be beneficial.

This work will be supervised by Dr. Vincenzo De Florio (mentor), Dr. Ahmed Nagy and Dr. Matteo Greco (co-mentors).