Projects

Professional / Research

KiMeKo: Promoting AI in Medical Technology

A collaborative project of leading AI institutes in Northern Germany, in partnership with top-tier medical technology companies and university hospitals. Leading research on extracting causal information from medical texts (e.g., guidelines, reports) and aligning AI-based expert systems with medical knowledge, bridging the gap between LLM-based AI technologies and real-world medical applications.

NLPLLMsMedical AICausal Extraction

Deep Learning for Interferometric Scattering (iSCAT) Microscopy

A project on deep learning for iSCAT microscopy focusing on segmentation, particle tracking, synthetic particle generation, and nanometer-scale particle size estimation. Since particle diameter is a strong biomarker for cancer and other diseases, this approach enables rapid, high-precision analysis of nanoscale biological changes, supporting early diagnosis and disease monitoring.

Computer VisionDeep LearningMedical ImagingPyTorch

Beam Pattern Reinforcement Learning

Reinforcement learning algorithms to optimize adaptive beamforming in mmWave and THz MIMO systems. Published in IEEE Transactions on Communications.

PythonPyTorchReinforcement Learning6G/THz

Photo Enhancer App

A web application featuring manual and RL-based editing with color histograms for photographers. Achieved state-of-the-art RL methods in tonal adjustment on the MIT-Adobe FiveK dataset.

Computer VisionReinforcement LearningDockerPython

Personal / Open Source

ModulaRL: Reinforcement Learning Library

An open-source, modular reinforcement learning library designed for rapid prototyping and reproducible experimentation. Built with PyTorch and designed for extensibility.

PythonPyTorchReinforcement LearningOpen Source

NutriMind

A personalized diet recommendation web application orchestrating front- and back-end services. Reached over 20,000 unique users and earned 250+ GitHub stars.

Scikit-LearnFastAPIStreamlitDocker ComposeMachine Learning

Emergent Flocking with Reinforcement Learning

Trains a Deep Double DQN agent to exhibit emergent flocking behavior (cohesion, separation and alignment).

PythonPyTorchReinforcement LearningMulti-Agent

AI Portfolio Chatbot

An interactive RAG chatbot that answers questions about my professional background. Built with a FAISS vector index, SentenceTransformers embeddings, and a Groq/Gemini LLM with streaming responses.

Next.jsFastAPIRAGFAISSDocker

pyBanania

A faithful Pygame recreation of the classic 1992 Windows 3.x game "Banania". Includes full game logic, sprite handling, and PyInstaller executable generation. Currently working on an RL agent that solves all levels of the game.

PythonPygameGame Development