Aleksandr Loginov

- Led four developers as Tech Lead for PD Predict, integrating ML models and APIs and launching the risk analytics module for 20+ analysts and external B2B clients. - Automated ingestion of 10M+ financial records from SPARK-Interfax API/XML sources, reducing data-mart preparation time by approximately 67%. - Improved a corporate PD model's Gini from 0.52 to 0.86 using tree-based ensembles and feature engineering; used SHAP to explain prediction drivers to stakeholders.

• Participated in designing the core calculation engine for an oil and gas pipeline route optimization system. • Applied graph algorithms (shortest paths, routing, min-cost flow); prototyped solutions in Python (NetworkX). • Researched the applicability of Graph Neural Networks for graph-based tasks.

CAE/MSC Adams, kinematic analysis.

Preparation of computational models, thermodynamic analysis, acoustic/mechanical calculations in ANSYS

Thesis: "LLM-based Intelligent Assistant for Optimizing Manufacturing Processes"