Artur Pelcharskyi

Center for Responsible AI at NYU · 1 yr

Developed geometry-aware loss functions and inter-atomic-distance evaluation metrics for molecular-docking generative models, using PyTorch for DiffGui/PMODIFF integration. Investigated structural constraints to improve chemical validity and docking quality. Created reusable preprocessing scripts computing statistical potentials from arbitrary datasets, including CrossDocked2020.

UCU Faculty of Applied Sciences / APPS UCU · Lviv, Ukraine · 2 yrs 6 mos

Assisted Calculus, Algorithms and Data Structures, Linear Algebra and Mathematical Methods of Machine Learning courses. Designed and maintained labs, graded work, provided academic support, led seminars/consultations and mentored course/project work.

AGROSEM · 6 mos

Developed an ML inventory-planning system for 30,000+ products, incorporating historical demand and budget constraints. Compared forecasting approaches for replenishment recommendations and built a web app with forecast visualisation and decision-support dashboards. Self-reported result: planning effort reduced from roughly one week by three employees to under one hour for a single operator.

Ukrainian Catholic University ML Lab · Lviv, Ukraine · 4 mos

Led a team of eight students researching optimal signal recovery from incomplete frequency-domain observations. Applied recovery methods to time-series imputation and benchmarked performance across varying missing-frequency-information levels.