Master Thesis - Optimizing Sparse 3D Occupancy Prediction for Embedded Devices - REF5666W

The following tasks are to be processed in this Master's thesis: You will work on sparsifying, pruning and quantizing CNN and transformer-based architectures for autonomous driving Design, implementation, and measured models and training algorithms for runtime optimized 3D semantic occupancy predictions Summarize research findings in a paper/ master thesis and/or patent submission

We are looking for you starting in February for a period of 6 months to support us in our AI laboratory in Berlin. We will need you mostly on site in Berlin. But we understand that there are days when you'd rather work from home. Therefore, together we find a solution that suits us all - remote working hours are therefore possible in consultation with your supervisor. The following tasks are to be processed in this Master's thesis: You will work on sparsifying, pruning and quantizing CNN and transformer-based architectures for autonomous driving Design, implementation, and measured models and training algorithms for runtime optimized 3D semantic occupancy predictions Summarize research findings in a paper/ master thesis and/or patent submission While not mandatory, this position is particularly well-suited for students who are considering a future in research or pursuing a PhD.

Are you wondering whether you are suitable for the thesis in the field of lightweight 3D Occupancy Prediction? Here are key qualifications we are looking for: You study in the field of Artificial Intelligence, Machine Learning, Computer Vision, Computer Science, Robotics, Mathematics or comparable courses of study, so that you have basic aspects of our area of responsibility Strong programming skills in Python are absolutely necessary in the context of this position You should have a strong interest in working on runtime optimization It would be great if you have experience in 3D perception, previous exposure to multihead attention, experience with deep learning, training and optimizing large architectures as well as already worked with Docker, PyTorch, OOP and Git Demonstrated research/software engineering experience through previous internships, work experience, coding competitions and/or research projects and papers are a plus Fluent English language skills in word and writing are absolutely necessary for us