If you are interested in working with an awesome team on the bleeding edge of AutoML, join us! For interested students, we have a separate page here. For all others, please find our current job opportunities below and join us ...
Need Additional Reasons to Join us?
Frank’s ERC Consolidator grant (and his previous 2 ERC grants). This grant of 2 million Euros (roughly $2 million US) allows us to do basic research on the foundations of the next generation of deep learning, no strings attached (other than doing excellent science). Please see this blog post on Deep Learning 2.0 for details about the project. In general, ERC grants are Europe’s most prestigious funding instrument, focused purely on an excellence-based approach to science. As stated in a recent Nature article, the ERC has helped Europe to “surpass the United States in terms of the most-cited scientific publications” and is “recognized as the best in the world in the way it supports fundamental research”.
The ML Freiburg group. ML Freiburg is amongst the leading groups in AutoML worldwide, with a focus on automated deep learning. This includes meta-learning (e.g., with 10 papers at the NeurIPS meta-learning workshop), efficient neural architecture search, efficient hyperparameter optimization, deep learning for tabular data, and automated algorithm design using ML in general. The group has won the first two international AutoML challenges (2015-2016 and 2017-2018), with continuously improving versions of its widely-used open-source tool Auto-sklearn, is also building the automated deep learning tool Auto-PyTorch and proposed the radically different approach of prior-data-fitted networks, which learned to do tabular classification in a single forward pass while approximating Bayesian inference. Frank is the general chair of the AutoML conference, after co-organizing the AutoML workshop series at ICML for eight years in a row. He also co-started the workshop series on Bayesian optimization at NeurIPS (since 2011), meta-learning at NeurIPS (since 2017), neural architecture search at ICLR (since 2020) and co-edited the AutoML book. The group developed the best available out-of-the-box tool for efficient hyperparameter optimization of neural networks and is amongst the world’s leading groups in neural architecture search and various other meta-algorithmic problems, such as algorithm configuration and algorithm selection, which have led to world championship titles in SAT solving and AI planning.
Resources and collaboration opportunities. ML Freiburg owns several large compute clusters, comprising about 300 GPUs and 1500 CPU cores. It also has (non-exclusive) access to a central cluster with 15.000 CPU cores. The group collaborates closely with the Bosch Center for Artificial Intelligence, which funds basic research on AutoML in the group with 6.4 million euros over 4 years. We also collaborate closely with the other AI groups in Freiburg, especially Thomas Brox’ computer vision group (leading to several new advances in neural architecture search), Tonio Ball’s neuromedical AI lab (leading to the first published work achieving state-of-the-art performance in EEG decoding with deep learning), the robotics group, Joschka Boedecker’s neurorobotics group, Abhinav Valada’s robot learning group and Josif Grabocka’s representation learning group. All of these groups are mostly working on deep learning and reinforcement learning these days, leading to exciting convergences and synergies. Frank is also Chief Expert AutoML at the Bosch Center for Artificial Intelligence. Finally, Frank is the director of the ELLIS unit Freiburg (one of the founding 17 units), and Frank is a fellow in the ELLIS programme on robust machine learning, opening up many opportunities for collaboration with other machine learning hot spots in Europe, such as the university of Oxford. The group also has close ties to many researchers in top industrial labs, including DeepMind, Google, Microsoft, Meta, Bosch and Samsung.
The University of Freiburg. Founded in 1457, the University of Freiburg is one of the oldest German universities and is now one of the nation’s leading research and teaching institutions, evidenced amongst others by it being one of the 23 members of the League of European Research Universities (LERU). It actively fosters interdisciplinary research (e.g., the Excellence cluster BrainLinks-BrainTools, which, amongst others, researches on deep learning for neuroscience), and it is one of the few universities offering world class research environments in the classical as well as in the modern disciplines. More than 24,000 students from over 100 nations are studying in 180 degree programs at 11 faculties. The university also successfully attracted the highest third-party funding per-professor in all of Germany.
The University of Freiburg is one of the 17 first excellent European universities that have been distinguished as ELLIS units for their excellence in machine learning. The university has a particularly strong group of faculty members in the field of artificial intelligence (including 3 ERC grant holders) covering a wide range of modern AI topics, such as machine learning, computer vision, robotics, planning, and control. The yearly budget in AI is more than 3M Euro and funds more than 50 PhD students. A robotics school and a deep learning compute cluster have been fostering the interdisciplinary research among AI researchers via the overarching methodology of deep learning. There are strong international relationships to other scientists and companies with an adjunct professor from DeepMind and three faculty members working
part time for Bosch, Toyota, and Amazon, respectively, and a faculty member advising the German parliament on AI. The AI research at the University transfers its results actively to other disciplines, such as the life sciences, sustainable systems engineering, and ethics. The AI researchers are involved in all four clusters of excellence as set up by the German government, one of these clusters being coordinated by a faculty member in AI.