Master Thesis – AI Assistant for R Code Review

Autoliv’s Traffic Safety Research team uses detailed crash databases for research that requires reliable data analysis. Researchers currently check the R code manually, a time-consuming approach that can still leave errors undetected. This thesis will investigate whether an AI-based reviewer can improve the quality and efficiency of these checks by considering the research question, analytical methods and crash-data domain knowledge.

The assignment

• Develop a working AI assistant that reviews existing R code rather than generating new code, checking logic, joins, filters, calculations, missing values and unexpected analytical results.
• Define a structured specification for analyses and a classification of coding errors.
• Create a labelled benchmark of R scripts, using synthetic data and deliberately inserted errors for development and testing.
• Investigate ways to incorporate database codebooks, coding standards and relevant literature, potentially through retrieval-augmented generation (RAG).
• Establish evaluation metrics and compare the assistant with manual review, assessing error detection and reductions in review effort.
• Account for secure handling of proprietary code and sensitive crash data, using anonymized or synthetic data where appropriate.
• Document the solution’s strengths and limitations and deliver a prototype, thesis report and final presentation.

Who we are looking for

• One or two Master’s students interested in combining software development, data analysis and AI.
• Experience programming in both R and Python.
• Interest in artificial intelligence, machine learning and data analysis.
• Prior knowledge of AI/ML, large language models or RAG is an advantage.
• A curious, analytical approach and motivation to apply emerging AI methods to practical research problems.

What we offer

• An opportunity to apply AI and software development to a real traffic safety research problem.
• Experience exploring how modern AI methods can be turned into a practical industrial application.

Practical details

Location: Vårgårda. The project is intended for 1–2 Master’s students. Start date, duration and application deadline are not specified. For further information, contact Hanna Jeppsson (hanna.jeppsson@autoliv.com), Junaid Shaikh (junaid.shaikh@autoliv.com), Jawwad Ahmed (jawwad.ahmed@autoliv.com) or Jordanka Kovaceva (jordanka.kovaceva@chalmers.se).