I prototyped a decentralized wireless federated learning setup for audio-based keyword spotting on Arduino (Portenta H7), then migrated the framework to Zephyr RTOS, integrated wireless communication (CoAP/Thread), and evaluated the system across accuracy, latency, and energy consumption. This work strengthened my ability to translate algorithms into robust, measurable embedded implementations and to iterate quickly from prototype to production-oriented software.
i built an anomaly detection for a RISC?V?based SoC intended for FPGA edge devices. I contributed to data collection and preprocessing from simulated attack scenarios and supported the implementation and optimization of trained models for resource-constrained embedded environments.
During my Masters in Electrical Engineering and Information Technology at TU Braunschweig , I deliberately combined machine-learning focused coursework/projects (e.g., pattern recognition and deep learning) with modules and hands-on labs in embedded systems, digital design, and computer architecture (e.g., FPGA/VHDL and RISC-V-related work).?
I concluded the degree with a Master?s thesis that brings both tracks together: , where I developed and evaluated a decentralized federated learning system for real-time speech recognition on resource-constrained embedded devices.
I prototyped a decentralized wireless federated learning setup for audio-based keyword spotting on Arduino (Portenta H7), then migrated the framework to Zephyr RTOS, integrated wireless communication (CoAP/Thread), and evaluated the system across accuracy, latency, and energy consumption. This work strengthened my ability to translate algorithms into robust, measurable embedded implementations and to iterate quickly from prototype to production-oriented software.
i built an anomaly detection for a RISC?V?based SoC intended for FPGA edge devices. I contributed to data collection and preprocessing from simulated attack scenarios and supported the implementation and optimization of trained models for resource-constrained embedded environments.
During my Masters in Electrical Engineering and Information Technology at TU Braunschweig , I deliberately combined machine-learning focused coursework/projects (e.g., pattern recognition and deep learning) with modules and hands-on labs in embedded systems, digital design, and computer architecture (e.g., FPGA/VHDL and RISC-V-related work).?
I concluded the degree with a Master?s thesis that brings both tracks together: , where I developed and evaluated a decentralized federated learning system for real-time speech recognition on resource-constrained embedded devices.