We are looking for a motivated student to support our AI research on-site in Freiburg. You will explore and improve denoising diffusion models for multimodal anomaly detection focusing on modern transformer architectures for image encoding and denoising. You will experiment with real-life industrial datasets as well as large publicly available benchmarks.
What are your tasks?
* Research and evaluate diffusion models for anomaly detection
* Work with transformer-based encoders (e.g., JIT) and diffusion transformers (DiT) denoisers
* Perform experiments on datasets such as KAPUTT
* Investigate recent advances (flow matching, sampling strategies, feature fusion)
* Document and present your results
This is what makes you special
* You are studying Computer Science, Artificial Intelligence, Data Science, Applied Mathematics or related.
* You are curious, proactive in solving complex research problems and enjoy experimenting with new ideas.
* You possess strong analytical thinking and enthusiasm for modern machine learning research.
* You already have some experience with generative models (e.g., diffusion models, VAEs, GANs) or anomaly detection methods.
* You have some initial knowledge Python, PyTorch or TensorFlow, Linux environments, GPUaccelerated trainings, and common ML tooling (Git, MLflow, etc.).
* You speak fluent English.
What you can expect from us
* Professional and personal support from experienced supervisors
* The opportunity to work independently on challenging projects
* A state-of-the-art working environment in an internationally active family business
* "State of the art" equipment to get off to a flying start
* Home office in consultation
* Individual company and ceam events
* Deutschlandticket is included!