Intern/Thesis/Working Student (f/m/d) - Evaluating and Improving LLM-based SE Solutions in SAP HANA

We help the world run better

At SAP, we keep it simple: you bring your best to us, and we'll bring out the best in you. We're builders touching over 20 industries and 80% of global commerce, and we need your unique talents to help shape what's next. The work is challenging – but it matters. You'll find a place where you can be yourself, prioritize your wellbeing, and truly belong. What's in it for you? Constant learning, skill growth, great benefits, and a team that wants you to grow and succeed.

What you'll build

Large language model (LLM) have a huge impact on software engineering. The context of SAP HANA is rather unique: A software project with over 10 million lines code, but not contained in the training data of any public LLM. We found that many results from public research about LLM usage in software engineering do not translate to SAP HANA.

This student position supports an internal research project driven by a PhD to evaluate the usefulness of LLM in the context of SAP HANA. The goal is the determine concrete recommendations of how to apply LLM in the context of SAP HANA for various tasks.

Some example tasks: Test generation, fault localization, LLM-as-a-judge, software reliability engineering, incident analysis, health checks, software engineering task automation, code reviews, static code analysis automation, bug fix automation, bug fix analysis, performance analysis and improvements, cost optimization, test flakiness improvements, benchmarking.

The student will have access to recent LLM models and also to GPU-based hardware for executing ML workloads (inferencing, learning, and so on).

We support both thesis and working student positions, details will be discussed and aligned in a first interview.

What you bring

Expected qualifications:

1. Student of computer science, data science, computer linguistics, mathematics or related field at a higher education institution
2. Experience with programming, machine learning and LLMs, demonstrated by existing projects.