Applied optimization Research tooling behind two peer-reviewed papers

overview

This work spans two optimization studies at the University of Prishtina: automated slideshow design using a hybrid Genetic Algorithm and Iterated Local Search approach, and traffic-signal scheduling using Iterated Local Search. I contributed as a co-author and built tooling for generating reproducible problem instances and validating candidate solutions.

problem

Optimization results are only credible when experiments can be repeated across controlled instances and candidate solutions can be checked against every constraint. Realistic benchmark generation, feasibility validation, scoring, and comparable time budgets are foundational research infrastructure—not administrative extras.

approach

I built browser-accessible generators for configurable synthetic instances and a validation framework for assessing solution feasibility, constraint adherence, and performance. The tools supported experimental pipelines that compared metaheuristics against competition baselines and other solvers across both generated and established benchmark sets.

outcome

The slideshow study was published in SN Operations Research Forum in 2023 and reported new benchmark results on two instances while evaluating a generated 55-instance test set. The traffic study was published in Discover Applied Sciences in 2025 and reported top rankings on eight specific instances within a 30-minute run budget.

selected details

• Co-author of two peer-reviewed Springer Nature optimization papers
• Reusable generators for synthetic slideshow and signalized-network instances
• Validation tooling for feasibility, constraints, scoring, and reproducible comparisons
• Research spanning genetic algorithms, iterated local search, and benchmark design