Feasibility Study and Architectural Design for Integrating LLMs with Power BI for Vehicle Analysis

Feasibility Study and Architectural Design for Integrating LLMs with Power BI for Vehicle Performance Analysis
30 hp - Power BI and LLM integration

Introduction
This study aims to explore the technical feasibility and architectural design of integrating Large Language Models (LLMs) with Power BI to streamline vehicle performance analysis. By examining the challenges and opportunities of this integration, the thesis will lay the foundation for future AI-driven data analysis systems in the vehicle testing department. Provided with the possibility of integrating LLMs into our daily work we expect to drastically increase the efficiency and speed of wok during planning, testing, and validation of vehicles. 

Background 
Traditional vehicle performance analysis in Power BI involves manually processing large datasets. LLMs can assist by enabling natural language queries and automating some data interpretation tasks, potentially improving both efficiency and accuracy in insights generation.

Objective 
To evaluate the feasibility and design a scalable architecture for LLM-Power BI integration. The study will focus on data flow, security, and user interaction while developing a prototype for proof of concept. An important objective is though to investigate whether it is enough to use available tools and APIs for this purpose, or we need to develop our models from the scratch. A thorough analysis of both cases is necessary to understand the challenges, opportunities, and complications each presents..

Job description 

• Investigate Power BI’s integration frameworks and APIs.
• Design an architecture for seamless LLM integration into Power BI.
• Prototype and test a small-scale LLM integration within Power BI.
• Assess the system’s impact on workflow efficiency and data insights. 
• Presenting and reporting of the intermediate and final results.

Education/program/focus
Master’s in Data Science, AI, Machine Learning, or related fields with a focus on predictive analytics, data-driven decision-making
Having knowledge of vehicle performance is a bonus.
Number of students: 1
Start date for the thesis work: 2025-02-03
Estimated duration: 22 weeks, until 2025-07-04. In average min. 2 office-days per week.
Language of work: English, Swedish.
Location of work: Södertälje and Stockholm, Sweden.
Students from outside Greater Stockholm can rent a subsidized Scania apartment in Södertälje.
Payment before taxes: 20'000 SEK at the start, 20'000 SEK after the final report was approved.

Contact persons and supervisors
Mohammad Davari, Data Analyst, Vehicle Validation, mohammad.davari@scania.com
Mattias Borg, Group manager, Vehicle Validation, mattias.borg@scania.com 

Application:
Please enclose personal letter, CV and the list of grades from the latest M.Sc. studies.
An academic reference is required. 

A background check might be conducted for this position. We are conducting interviews continuously and may close the recruitment earlier than the date specified.