TR
BlogProjectsAbout
Projects

Mobile · Biomedical · Applied AI

I turn biomedical research and complex mobile constraints into products that can be measured, shipped and used—from heartbeat identity and safety systems to public healthcare apps and private on-device AI.
Flutter
Kotlin
Python
Production healthcare apps
I joined Siemens Healthineers as the sole Flutter developer on a new mobile platform, established its application and delivery foundations, and helped it grow into three public apps now built by a full Scrum team. My contributions span architecture, native interoperability, accessibility and delivery.
Applied research
A current example of how I evaluate an idea across model quality, product constraints and real devices.

On-device medical AI

Private AI for medical-result interpretation
A three-pipeline evaluation of structured extraction and interpretation, followed by matched Android and iOS benchmarks. The goal was not a demo—it was to understand what a private, constrained assistant could realistically do on-device.
20
small models
118
reviewed cases
95%
best partial match
Selected systems
Projects with substantial research, engineering or real-world validation behind them.
ECG biometric authentication mobile application

Published biomedical-AI research

ECG Biometric Authentication
Identity verification from a single heartbeat, combining ECG signal processing, interpretable neural networks and practical on-device inference.
Python
Kotlin
Android
TensorFlow
TFLite
98.9%
average TAR
80
participants
99.5%
average TRR
BeSafeBox fall and car-crash detection application

Recognized bachelor’s thesis

BeSafeBox
A modular Android system for fall and car-crash detection, developed from data collection through model validation and reliable background execution.
Java
Python
Android
TFLite
86–90%
fall accuracy
520+ h
field testing
0
false crash alarms
SensorBox Android sensor-recording application

Mobile and wearable sensing

SensorBox
Android and Wear OS sensor capture with annotations, alarms and multi-API subscriptions, built around a shared sensor-centralization module.
Kotlin
Android
Wear OS
MVI
Raw data
sensor capture
Wear OS
wearable support
Notes
annotations and timers
Open source
Smaller, focused tools for interoperability, biomedical signals and Android development.
flutter_interop_fhir
Reads Android Health Connect records into FHIR-oriented Dart models through JNI and jnigen instead of platform channels.
Flutter
Android
FHIR
JNI
ECG R-wave detector diagram for Kotlin
Kotlin ECG Detectors
Online and offline implementations of widely used ECG R-wave detectors for Kotlin applications.
Kotlin
ECG
ECG R-wave detector diagram for C++
C++ ECG Detectors
Portable C++ implementations of the same detector family for native and embedded integrations.
C++
ECG
Interpretability analysis of a one-dimensional ECG model
SignalScreen
Occlusion, saliency and Grad-CAM interpretation tools for one-dimensional TensorFlow models.
Python
TensorFlow
BluModify Android application
BluModify
A focused Android utility that warns when Bluetooth is enabled without an active connected device.
Kotlin
Android

The rest of the work

Experiments, examples and teaching repositories

My GitHub includes mobile utilities, architecture examples, RAG experiments, agent tooling and supporting research code.