Computational Chemist (Eni S.p.A.)
Fine-tuned ML interatomic-potential models to improve agreement with DFT reference data and defined numerical targets for training evaluation. Built version-controlled Python training and simulation pipelines on Linux/HPC using reproducible inputs and convergence checks. Analyzed where models succeeded or failed against reference calculations and refined workflows iteratively. • Set numerical targets for DFT agreement and evaluation • Implemented convergence checks and reproducible training pipelines • Conducted model performance assessment against reference calculations • Iteratively improved training workflow based on failure modes