This fellowship aims at investigating deep learning models for passive acoustic monitoring of animal species. The data will primarily be acoustic data. The fellow will research state-of-the-art techniques in deep learning and mathematics to address critical conservation problems. The fellow will develop software code for audio processing (typically in Python), and suitable deep neural networks, typically in TensorFlow or other libraries.
The fellow will work closely with Dr. Emmanuel Dufourq and the Machine Learning for Ecology Research Group. The fellow will work at AIMS South Africa which is located in the beautiful city of Muizenberg, Cape Town, South Africa. AIMS is located just a few meters from Muizenberg beach!
The fellow will also have the opportunity to travel to Rwanda in both years, to visit AIMS Rwanda, engage with the researchers there, and have fun along the way.
In reality this position is a little more "computer sciencey" and a little less "field work". However, if the candidate loves field work, then they are free to do so to their heart's content. Field work in Rwanda would probably be really cool and fun. Cape Town offers a number of really awesome field work locations, however, I (Emmanuel) tend to be behind the computer :) In practice, you'd have a lot of freedom to explore your own cool ideas related to ML for bioacoustics and conservation tech in general. You would have access to strong cohorts of Google DeepMind funded Masters students in South Africa, should you wish to supervise and build your academic portfolio. You'd also have access to students from a large number of African countries, so your impact would be massive from a supervision perspective.
Since this is a postdoc position in South Africa, the candidate will not pay tax on this funding, as per the SARS rules for postdocs.
If you're excited to help with animal conservation and use deep learning to achieve this, then you should apply! If you love to code and chat about research, then you'll have a great time!
Deadline: I have put the 31st January 2024, but if you need a little more time, then please reach out to me directly (email in PDF)!