Ömer Ekmekcioğlu

Warwick Business School

University of Warwick

Coventry CV4 7AL, UK

I am a Ph.D. candidate at Warwick Business School, University of Warwick, advised by Juergen Branke and Nursen Aydin. My research is on expensive decision-making problems. Given a fixed budget for expensive evaluations, the goal is to choose the next experiment well while learning a black-box objective along the way. I work on Bayesian optimization and extend it to problem structures that are complex and difficult to solve.

Most of my recent work is on bilevel problems. I have also worked on high-dimensional and multi-objective problems, and on surrogate models.

My current work brings principled Bayesian optimization methods to automated discovery with large language models.

Alongside my Ph.D., I have been a research assistant at the UCL School of Management since 2023, where I built a decision-support tool for thalassemia treatment planning that is used by physicians. Before Warwick, I received an M.S. in Industrial Engineering and a B.S. in Electrical and Electronics Engineering from Bilkent University.

I am on the 2026–27 job market for faculty positions in operations research, operations management and business analytics, and for research roles in industry. My CV is available here.

Job Market Paper

Bayesian Optimization of Bilevel Problems

Ömer Ekmekcioğlu, Nursen Aydin, and Juergen Branke

2024

Working paper. Presented at the SIAM Conference on Optimization (OP26), Edinburgh, June 2026.

Abs arXiv

Bilevel optimization, a hierarchical mathematical framework where one optimization problem is nested within another, has emerged as a powerful tool for modeling complex decision-making processes in various fields such as economics, engineering, and machine learning. This paper focuses on bilevel optimization where both upper-level and lower-level functions are black boxes and expensive to evaluate. We propose a Bayesian Optimization framework that models the upper and lower-level functions as Gaussian processes over the combined space of upper and lower-level decisions, allowing us to exploit knowledge transfer between different sub-problems. Additionally, we propose a novel acquisition function for this model. Our experimental results demonstrate that the proposed algorithm is highly sample-efficient and outperforms existing methods in finding high-quality solutions.

News

Jun 2026	Presented my job market paper at the SIAM Conference on Optimization (OP26), Edinburgh.
Jun 2026	Presented Combining Gaussian Processes and Neural Networks as Surrogate Models in Bayesian Optimization at LION 20, Milan; LNCS proceedings forthcoming.

Selected Publications

Combining Gaussian Processes and Neural Networks as Surrogate Models in Bayesian Optimization

Ömer Ekmekcioğlu, Juergen Branke, and Nursen Aydin

In Learning and Intelligent Optimization (LION 20), Lecture Notes in Computer Science, 2026

Accepted, forthcoming.
Graph Neural Networks for Deep Portfolio Optimization

Ömer Ekmekcioğlu and Mustafa Ç. Pınar

Neural Computing and Applications, 2023

DOI Code

The full list of papers is on the Research page.

References

Juergen Branke (advisor)
Warwick Business School
University of Warwick
juergen.branke@wbs.ac.uk

Nursen Aydin (advisor)
Warwick Business School
University of Warwick
nursen.aydin@wbs.ac.uk

Mustafa Ç. Pınar
Industrial Engineering
Bilkent University
mustafap@bilkent.edu.tr

Ersin Korpeoglu
UCL School of Management
University College London
e.korpeoglu@ucl.ac.uk

Kenan Arifoglu
UCL School of Management
University College London
k.arifoglu@ucl.ac.uk