Bridging Robotics, AI, and Healthcare.
Systems engineer and researcher building tools that translate muscle signals into real-world control, from wearable hardware and embedded firmware to deployed machine learning pipelines.
32-channel wearable · 92.9% accuracy · 14 participants
Erlangen, Germany
About
M.Sc. in Medical Robotics at FAU Erlangen-Nurnberg (GPA 1.2), combining biosignal engineering with a track record of shipping production systems. Eight years as the sole engineer of a commercial IoT platform. Three years of hands-on research across three FAU labs covering wearable biosignal acquisition, ultrasound-based assistive robotics, and ROS-based systems.
Capabilities
Selected Work
92.9% accuracy · 14 participants (7F, 7M) · 24 s protocol
Prototyped and validated an end-to-end HD-sEMG gesture-recognition system with real-time inference and a live two-player game. First simultaneous multi-device OTB acquisition in the research group.
200+ buildings · 8 years · zero security incidents
Three hardware generations from AVR/GSM to full cloud IoT with 433 MHz RF wireless gate control. Every layer owned by one engineer: custom PCBs, embedded firmware, MQTT, REST API, MySQL, and PWA frontend.
MOSAIC
IEEE Access 2025
A-Mode · B-Mode Ultrasound
Fraunhofer SDK · C#
Contribution to MOSAIC · IEEE Access 2025
First A/B-mode ultrasound acquisition pipeline integrated into MOSAIC. Thread-isolated C# pipeline, per-sample timestamps, low-memory recording mode, and gel-material evaluation via confusion matrices.
9.6% task time reduction · p = 0.001 · 17 participants
GDPR-compliant 3-stage voice pipeline: Vosk offline ASR, Levenshtein fuzzy matching, Metaphone phonetic fallback, LLM last resort. Sub-100 ms response. Within-subject study with NASA-TLX.
32-channel EMG · MIDI kinematics · motor control research
Dedicated Piano Tab inside MyoGestic: simultaneous 32-channel HD-sEMG and MIDI recording. Automatic trigger after three consecutive correct performances ensures ground-truth quality datasets.
ROS2 Jazzy on NVIDIA Jetson · SLAM navigation · 2.5 years
Navigation and perception for a mobile social robot: SLAM, Z-camera depth, and full ROS2 integration. Deployed ROS2 Jazzy Jalisco entirely from source on NVIDIA Jetson. Tutored the Robotic Frameworks course.
63 variables reduced to 15 · 18 s training · real-time feasible
Geometric reformulation reducing HD-sEMG-to-kinematics regression from 63 three-dimensional variables to 15 independent linear regression tasks. One best-matched EMG channel per joint angle.
Custom animation engine · visual class editor · JSON-driven sessions
Server-based interactive learning platform with a shared animation engine powering both a visual class creator and a student-facing mobile frontend. Synchronised video playback with interactive question types.
Demos
Simultaneous HD-sEMG and MIDI recording · n-squared Lab FAU
Real-time RPS classification · OT Bioelettronica Muovi 32-ch
HD-sEMG gamification study · FAU Erlangen-Nurnberg
A working day at FAPS Lab · robotics research in action
Get in touch
Open to positions in Germany. Wearable systems, biosignal engineering, medical robotics.
Erlangen, Germany · Open to Munich