The research field of detecting malicious android apps has been analyzed in various research papers. Besides recent deep learning methods, solutions with classical machine learning approaches have been put to debate. However, an open-source framework where all proposed extracted features to train machine learning classifiers are accumulated is missing yet. This includes embedded features of deep learning models. Such a framework can be used to revisit the engineered features and evaluate them against each other to discover the most informative ones. Also, this framework could be used by other researchers as a benchmarking tool to compare the performance of new proposed features with the state-of-the-art. 
The task of this thesis is to review (part of) the literature regarding android malware detection and build a (comprehensive) android malware detection framework. Depending on the students' capability, a feature analysis component could be added as well. 

Although Python is the preferred programming language, the utilized programming language can vary based on the student's knowledge and preferences.

References:

1. Yes, Machine Learning Can Be More Secure! A Case Study on Android Malware Detection, 2017
2. A Review on The Use of Deep Learning in Android Malware Detection, 2018

3. Automatic Feature Engineering: Learning to Detect Malware by Mining the Scientific Literature

4. Android Malware Detection: A Survey, 2018