What's Happening?
SAP is offering an internship, thesis, or working student position focused on evaluating and improving Large Language Model (LLM)-based software engineering solutions within the context of SAP HANA. This role is part of an internal research project driven
by a PhD candidate, aiming to determine concrete recommendations for applying LLMs to various tasks within SAP HANA. The SAP HANA project is unique due to its extensive codebase of over 10 million lines, which is not included in the training data of public LLMs. Consequently, many findings from public research on LLM usage in software engineering do not directly translate to SAP HANA. The intern will have access to recent LLM models and GPU-based hardware for machine learning workloads, including inferencing and learning. Example tasks for the intern include test generation, fault localization, LLM-as-a-judge applications, software reliability engineering, incident analysis, health checks, software engineering task automation, code reviews, static code analysis automation, bug fix automation and analysis, performance analysis and improvements, cost optimization, test flakiness improvements, and benchmarking. The position requires a student of computer science, data science, computer linguistics, mathematics, or a related field with experience in programming, machine learning, and LLMs, demonstrated through existing projects. A solid foundation in machine learning and statistics, along with practical application of LLMs, is also required. The role is based in Walldorf, Germany, with some home office days negotiable.
Why It's Important?
This internship highlights SAP's proactive approach to integrating cutting-edge artificial intelligence, specifically Large Language Models, into its core enterprise resource planning (ERP) software, SAP HANA. Given SAP's significant global presence, with over 400,000 customers and 200 million users worldwide, the successful application of LLMs could lead to substantial improvements in software development efficiency, reliability, and cost-effectiveness across numerous industries. The unique challenge of adapting LLM research to SAP HANA's proprietary and extensive codebase underscores the need for specialized research and development. If successful, this initiative could set new industry standards for how large-scale, complex software systems leverage AI for engineering tasks. The findings from this research could influence future software development methodologies, potentially leading to faster innovation cycles, reduced bug rates, and optimized performance for businesses relying on SAP's solutions. This move also signals SAP's commitment to staying at the forefront of technological advancements, ensuring its offerings remain competitive and relevant in an increasingly AI-driven landscape.
What's Next?
The immediate next step involves the selection and onboarding of a qualified student for this internship or thesis position. Following this, the student, under the guidance of the PhD candidate, will commence the evaluation and improvement of LLM-based software engineering solutions for SAP HANA. The project aims to generate concrete recommendations for applying LLMs to various software engineering tasks. The outcomes of this research could lead to the development of new AI-powered tools and features within SAP HANA, potentially enhancing its capabilities in areas like automated code generation, intelligent debugging, and predictive maintenance for software. Furthermore, the insights gained from this project could inform SAP's broader AI strategy, influencing how the company integrates LLMs into other products and services. The success of this initiative could also spur further academic and industry collaboration in adapting general LLM research to highly specialized and proprietary software environments, potentially leading to new methodologies and best practices in AI-driven software engineering.
Beyond the Headlines
The deeper implications of SAP's investment in LLM research for SAP HANA extend to the evolving landscape of software engineering and the future of enterprise technology. The challenge of applying public LLM research to a proprietary system like SAP HANA, which is not part of public training data, highlights a critical hurdle for many large enterprises. This effort could pioneer methods for fine-tuning or adapting general-purpose LLMs to specific, domain-rich, and often confidential corporate datasets. Success in this area could lead to a paradigm shift in how companies manage and evolve their legacy systems, potentially unlocking significant value by automating complex development and maintenance tasks. Ethically, the project will need to address concerns around data privacy and intellectual property when using LLMs, especially in a corporate context. The long-term shift could see a greater reliance on AI for critical software functions, raising questions about human oversight, accountability, and the potential for AI-induced biases in enterprise systems. This initiative also underscores the growing demand for professionals skilled in both software engineering and advanced AI, signaling a future where these two fields are increasingly intertwined.
