Loris Gjini
''Curiosity is the engine of achievement''
About Me
Hi, I'm Loris.
By day I wrangle data and write SQL queries. By night I'm lost in a book, a habit that's been with me since I was a kid. If you can't find me, I'm probably on the floor with my laptop—yes, actually on the floor. Something about it helps me think. My friends call it weird. I call it productive eccentricity.
When I get deep into a problem, I tend to pace around and talk to myself like a human fidget spinner. That same energy comes out when I'm discussing books or tossing around ideas with open-minded people who don't mind the occasional philosophical tangent.
I'm a little shy at first, but once that melts away I'm the friend who checks in regularly and genuinely wants to know how you're doing. I'll even watch TV series with you, though I secretly believe books are better—after all, you can control the pace.
Lately I've been on a bit of a self-discovery journey. Like most people, I sometimes find myself wondering where I fit in and what unique perspective I can bring to the table. I don't have it all figured out (who does?), but I'm starting to see that understanding yourself is a lot like reading a great mystery: you're both the detective and the main character.
My Projects
Real-world solutions in data engineering, machine learning, automation, and enterprise systems built during my professional experience.
Enterprise ETL Pipeline with Azure Data Factory
Comprehensive ETL ecosystem integrating multiple ERP systems including banking platforms, insurance policy systems, and external vendor feeds. Achieved 99.5% pipeline success rate, 80% reduction in manual processing, and 70% improvement in processing efficiency through intelligent scheduling and automated optimization.
Financial Delinquency Prediction System
End-to-end ML system for predicting loan defaults using ensemble methods (XGBoost, Random Forest) with time-based cross-validation. Achieved 85%+ precision, 78%+ recall, and 0.89 ROC-AUC through advanced feature engineering and hyperparameter optimization with Optuna.
Moody's API Integration & Ownership Analysis
Enterprise Python system integrating Moody's Orbis API for automated corporate ownership analysis using Italian P.IVA search. Features NetworkX graph-based calculations, Ultimate Beneficial Owner detection, circular ownership handling, and path tracing algorithms for regulatory compliance.
Pillar XML Parser
High-performance XML processing system with custom streaming algorithms for multi-gigabyte financial documents. Achieved 75% faster processing and 60% memory reduction through constant memory usage patterns while maintaining 99.9% data accuracy with comprehensive validation.
Smart Office Seating Algorithm
Mathematical optimization system using Gaussian distance scoring, social clustering, and Lucy penalty weighting for intelligent office seating assignments. Implements atomic database operations preventing race conditions with zero seat conflicts and 89% user satisfaction.
Rescue Maze Galilei Robot
Autonomous rescue robot for hazardous maze navigation using advanced pathfinding algorithms, computer vision, and multi-sensor fusion. Winner of Best Software Award at RoboCup Junior 2017 Nagoya, Japan with 95% victim detection rate and 8-minute maze completion time.
Skills & Expertise
Technologies and capabilities I've developed through professional experience and continuous learning.
Data Engineering
Programming
Cloud & Infrastructure
Analytics & Machine Learning
Professional Skills
Blog
Thoughts, tutorials, and insights about data science, technology, and my personal journey.
Coming Soon
My blog section is currently under development. I'm preparing articles about:
- Data analysis techniques and best practices
- Azure Data Factory workflows
- Machine learning applications in business
- Book recommendations and insights
I'm excited to share my knowledge and experiences with you. If you have specific topics you'd like me to cover, feel free to reach out!