Antoni Moszyński

Developed novel methods and implemented a Python-based image-analysis pipeline for medical images of esophageal cancer to extract extracellular matrix features. Developed a custom Wasserstein-distance metric module for ECM fiber-alignment measurement alongside standard image processing techniques.

Sole developer on a Flask-based REST API and React front end for the “Historical Avatars” exhibit engine, integrating with external databases and client systems. Containerized the application stack (API, front end, and PostgreSQL) via Docker Compose and deployed it on Oracle Cloud and OVH, handling monitoring and incident response. Liaised directly with client stakeholders to collect requirements, troubleshoot integration issues, and roll out updates.

Applied state of the art computer-vision methods to explore therapy-response prediction on histopathology images. Built and maintained reproducible preprocessing, training and evaluation scripts, ensuring consistent experiment tracking and ease of model re-runs using PyThorch Lightning framework. Presented model development progress and challenges to a multidisciplinary team of clinicians and bioinformaticians.

Built a custom Java framework for large-scale epidemic simulations (50M+ agents), enabling spatiotemporal Covid-19 forecasting for the Polish Ministry of Health. Optimized multi-threaded performance through spatial partitioning and conflict resolution logic. Engineered synthetic social-network generator using geospatial data.