• Worked on TensorDSE, a Design-Space Exploration framework to guide machine learning model deployments.

• Evaluated the performance of various ML models across GPUs, CPUs and TPUs with TensorFlow Lite.

• Generated cost analysis reports for Google’s Coral Edge TPU via USB traffic analysis (PyShark) during inference.

• TensorDSE used reports to accelerate/distribute a model’s inference optimally onto available hardware devices.

• Added unit-tests (GTest) and code coverage (lcov) to safety critical features of their motor’s embedded controller.

• Developed tooling for state simulations of their electric motor via Linux’s virtual CAN interface and mock APIs.

• Extended their firmware update system used by 20+ clients, consisting of partial updates via CAN bus.

• Automated build and testing workflows via Jenkinsfiles, Makefiles and CMake for a team of over 10 engineers.

• The ESPL course took place within the ECE Bachelor at TUM.

• I assisted students in their laboratory FreeRTOS projects, which consisted of multi-threaded applications in C designed for an embedded console.

• This went from from code review to software engineering practices and design.

Research internship at the EDA department within TUM, where I developed a USB analysis tool to trace the communication between Google Coral Edge's TPU and the host system during a machine learning model's inference.