Build the AI systems that take real work off real teams.

What you'll build

You'll own end-to-end AI features inside real customer systems. Not isolated notebooks — production pipelines that have to run every day, against messy data, in regulated industries.

• NLP pipelines for document understanding, classification and search.
• Image & vision models for quality inspection, OCR and document recognition.
• LLM-powered workflows: retrieval-augmented generation, agentic tooling, structured-output extraction.
• Predictive models for forecasting, anomaly detection and decision support.
• Model evaluation, monitoring and the boring-but-critical work of keeping production AI honest.

What we expect

• Solid foundations in machine learning and modern Python (PyTorch, Hugging Face, scikit-learn).
• Practical experience shipping NLP, computer vision or LLM-based features into production.
• Comfortable with web development — building APIs, integrating with frontends, designing data models.
• DevOps fluency: Docker, CI/CD, observability, EU-cloud deployment (AWS / Azure / Hetzner).
• You write code that others can read, debug and run six months from now.
• German or English — fluent in at least one, comfortable in the other.

What you'll get

• Real ownership of production AI systems — no babysitting demos.
• A short feedback loop with users and decision-makers, including the founders.
• Modern stack, EU infrastructure, sane tooling. No legacy graveyards.
• Flexible hybrid setup — Aachen as the home base.
• Time and budget to learn: conferences, courses, certifications.