At Centrum výzkumu Řež, I developed an end-to-end Python machine learning workflow for simulation data. My XGBoost model predicted dissipated energy with a held-out R² of 0.966 and sub-millisecond inference, compared with 2–6 hour finite-element simulations.
I deployed the models through a FastAPI and Streamlit application on Microsoft Azure, with GitHub Actions automating deployment. Previously, at Výzkumný a zkušební ústav, I investigated cold-spray and thermal-spray technologies and co-authored and secured funding for two international OFFERR/SNETP research projects with EDF France.

