For my bachelor’s thesis, I’ve worked on automatically recognizing body activities using smartphone- and smartwatch-sensors and machine learning. If you can read German and are interested in the topic, please feel free to read the whole thesis. To achieve my goal, besides developing the method itself, I’ve had to record movement data from ten subjects for ten activities each to evaluate the method. The complete anonymized dataset is available for download. It is encoded using the protobuf schema found in the archive.
For my master’s thesis, I’ve evaluated the performance of high-level programming languages in the context of High Performance Computing, i.e. supercomputing. More specifically, I’ve compared Java and Rust to C++ by developing microbenchmarks of constructs that are critical to parallel computing. These constructs were extracted from programs that represent real-world applications. Later on, I’ve ported the entirety of these programs from C++ to Rust and Java to also assess their real-world performance. The thesis is freely available here.