PhD – Generative Models for Closed-loop Synthesis

We are conducting cutting-edge research on advanced generative models aimed at enhancing data efficiency in Bosch systems. We are seeking a PhD student who is passionate about exploring innovative applications of generative models (such as diffusion and autoregressive models) to simulate real-world scenarios for AI training and validation.

The development of AI models is often an iterative process that requires increasingly large datasets to address long-tail cases that are not represented in existing data. However, collecting data from the real world can be time-consuming and expensive, hindering the automation of the data loop. The objective of this thesis is to create new methodologies that enable generative models to substitute for the real-world, facilitating closed-loop interactions. This may involve designing novel control mechanisms to efficiently sample the required data and respond to interactions.

As a member of our team, you will:

• Develop novel deep generative models (e.g., diffusion models) as data sources to enhance the training and validation of downstream models.
• Collaborate with experts in deep learning and computer vision at the Bosch Center for AI to brainstorm and develop new ideas.
• Aim for publications in top-tier journals and conferences.

• Education: excellent degree in Computer Science, or related field with focus on Computer Vision and Deep Learning
• Experience and Knowledge: strong background in deep learning and computer vision, experience with deep learning frameworks (TensorFlow, PyTorch, etc.), strong programming skills, in particular Python, knowledge and experience in deep generative modeling as well as foundation models are a plus, experience with publication of peer-reviewed research papers is beneficial
• Enthusiasm: motivation to work in an interdisciplinary and international team
• Languages: very good English skills and academic writing skills