Working Student (m/f/x) - LLM Infrastructure & MLOps Platform Engineering

The appliedAI Initiative elevates Europe’s industry to become shapers in the age of AI, creating a future that we desire to live in. Through our ecosystem of strong partners, comprehensive programs, services and solutions we advance companies holistically on their AI journey. With our uniquely diverse, interdisciplinary, and driven team, we build on deep technical expertise while specializing on applicability of the solutions. We share a mindset of collaboration and openness with our ecosystem of leading companies and tech players. Our common goal is to strengthen the European landscape, jointly creating unique knowledge for applying trustworthy AI. We believe that AI Agents will be the dominant topic for AI in the near future and that appliedAI with its partner ecosystem will impact how this technology is being used in industry and beyond in Europe.

appliedAI’s Internal AI team sits at the heart of how we bring AI into the real world — keeping the infrastructure that powers our internal tools and partner-facing AI products running reliably and at scale. As a working student (m/f/x), you’ll join a small team of experienced engineers who care deeply about operational quality and are happy to share what they know. This is a hands-on role where you’ll build practical intuition for production AI systems while contributing meaningfully from day one.

Your tasks will include:

• Support the operation of containerized model-serving systems (Docker Compose) across staging and GPU clusters (H100 and V100)
• Monitor system health, inspect logs, and help troubleshoot issues in collaboration with platform engineers
• Assist with deployment workflows using GitLab CI/CD, including staging validation and controlled rollouts
• Support the LiteLLM routing layer and help maintain model endpoints (chat, reasoning, embedding, speech APIs)
• Run API and integration tests to ensure OpenAI-compatible behavior across services
• Contribute to monitoring and observability by maintaining Grafana dashboards and Prometheus alerting rules