Passing a medical licensing exam is one thing. Supporting a real clinical decision with incomplete data, rare edge cases, and a physician waiting for an answer is another problem entirely.
Our research group AI for Women's Health is hiring student assistants to work on the VIOLET project, a nationally funded initiative developing a hybrid AI framework for guideline-based treatment decision support in gynecological oncology. The work combines knowledge graphs, retrieval-augmented generation (RAG), and LLM-based multi-agent systems and is grounded in real clinical data from the TUM University Hospital's data integration center.
Your job will be to help make these systems actually work in practice: probing where foundation models fail to follow clinical reasoning, building evaluation pipelines that go beyond standard benchmarks, and pushing inference to run on the kind of hardware that hospitals realistically deploy.
What we're looking for:
- Bachelor's or Master's student in Computer Science, Mathematics, Physics, or a related field
- Solid Python and PyTorch skills
- Genuine curiosity about the gap between model performance and clinical reliability
- Rigor in how you think, code, and document
- Prior experience in medical AI is welcome but not the deciding factor
What you get:
- Access to real, curated clinical datasets within a regulated research environment
- Substantial GPU infrastructure
- A short feedback loop between technical work and clinical application
- Both engineering and clinical mentorship within the team