Nico Loesch

• Developed scalable ETL pipelines for large-scale EHR analysis, integrating traditional NLP and LLMs for hybrid clinical data processing.
• Architected a RAG-based workflow to extract and ground clinical entities to standardised ontologies.
• Engineered CI/CD pipelines to automate the deployment of containerised ML applications.
• Leading the integration of LLM Agents using MCP and ACP to enable dynamic tool-use.

• Fine-tuning and prompt-engineering of LLMs for advanced social media text analysis, successfully enabling automated detection of cancer burden
• Semantic analysis, topic modelling and clustering of large-scale social media data using transformer-based language models

• Fundamentals of Data Analytics & Databases

• Developed software for robotic surgery in an agile development environment using CI/CD practices
• Performed validation on software implementations to ensure reliability and compliance with medical standards

• Developed a labelling GUI for medical image segmentation
• Developed real-time control algorithms for robotic arm manipulation, including gaze and head tracking using VR headsets

• Developed motion curves for an artificial 3D-printed cardiac phantom to simulate realistic cardiac motion
• Performed automatic segmentation of coronary artery calcification in CT images

• Weakly-supervised brain tumour segmentation of multi-modal MRI using diffusion models
• Multi-disciplinary research in collaboration with the Children's Hospital at Westmead, Sydney

thesis:Towards Robust Clinical Segmentation of Paediatric Brain Tumours in Magnetic Resonance Imaging Using Weakly-supervised Deep Neural Networks

• Erasmus exchange semester at Lund University, Sweden

thesis:Efficient Medical Image Segmentation with Graph Neural Networks 1.3/5.0

thesis:Development of a Protocol to Quantify Spatially-resolved Magnetic Resonance Spectroscopy in Neurology 1.6/5.0