Student Assistant: Build the Infrastructure Behind AI Safety Competitions

07.05.2026, Studentische Hilfskräfte, Praktikantenstellen, Studienarbeiten

Neural networks are deployed in safety-critical systems — from autonomous vehicles to medical diagnostics. How do we know they actually work? Every year, the international research community tests the best verification tools against each other in two major competitions:

- VNN-COMP: Website (https://vnn-comp.github.io/) | Submission system (https://vnn.repeatability.cps.cit.tum.de/)
- ARCH-COMP: Website (https://cps-vo.org/group/ARCH/FriendlyCompetition) | Submission system (https://arch.repeatability.cps.cit.tum.de/)

Both competitions rely on custom-built submission systems that manage benchmarks, execute verification tools on cloud infrastructure, collect results, and present them to the community. That's what you'll be working on.

Your tasks include:
- Full-stack web development to run the competitions
- Navigate spontaneous bugs in high-stakes situations
- Interest in related tasks

Essential skills:
- Web development (Django, REST APIs, React)
- Strong programming skills (across languages, mostly Python), and general software development (Linux, Git, Docker, Databases, CI/CD)
- Experience in server and cloud management (AWS) is a plus

We offer:
- Hands-on experience in international competitions
- Flexible working hours up to 20h/week
- Thorough onboarding with the previous HiWi to get you up to speed quickly

If you are interested, feel free to send your CV+transcript to: nico.holzinger@tum.de

Kontakt: nico.holzinger@tum.de