PhD - Foundation Models for Improving and Auditing Data at Scale

We conduct cutting-edge deep learning research with a focus on data. We are looking for a motivated PhD student eager to delve into research on deep learning methods for improving and auditing data, as well as to explore their real-world applications for further improving the performance, reliability, and efficiency of AI models.

The goal of this position is to develop novel and automated data auditing methods, leveraging foundation models. These methods shall enable data analysis at scale, provide insights into models' generalization capabilities and aid performance estimation in new domains. Additionally, outcomes from data auditing will be utilized to improve data quality, e.g., pruning less relevant samples, adding more valuable ones, and generating advanced annotations to expedite learning with smaller models and model evaluation.

• As a PhD student in our team, you will innovate and automate data auditing methodologies for analyzing data diversity, coverage as well as biases, leveraging deep generative models and foundation models.
• In addition, you will advance data filtering and annotation strategies to optimize training efficiency as well as streamline model evaluation processes.
• You collaborate with Machine Learning as well as Computer Vision experts and publish in top-tier journals as well as conferences.
• Furthermore, you will discuss and develop new ideas within the Deep Learning research team at Bosch Corporate Research (CR).

Requirements

• Education: excellent degree in Computer Science or related field with focus on Machine Learning/Deep Learning
• Experience and Knowledge: strong background and experience in Deep Learning, strong programming skills, in particular Python, knowledge in Deep Learning frameworks (TensorFlow, PyTorch, etc.), knowledge and experience in foundation models, generative models as well as out-of-distribution detection are a plus
• Enthusiasm: motivation to work in an interdisciplinary and international team
• Languages: fluent in English (written and spoken)