About Me

Hi! I am a PhD student in Computer Science at Saarland University as part of the CS@Max Planck Doctoral Program, and I conduct my research at the Max Planck Institute for Informatics. I completed my B.Sc. in Mathematics, with a minor in Statistics, at Middle East Technical University (METU), where I graduated with High Honors.

My research examines how learned representations support reasoning and sequential decision-making, and why model performance degrades when tasks require long-horizon behavior, composition of previously learned capabilities, or generalization beyond the training distribution. I am interested in studying these questions across foundation models, autonomous agents, and embodied systems.

During my previous Research Immersion Lab, I worked on self-supervised latent-action representations learned from human videos and their use in robot action generation. The project investigated whether compact representations of visual transitions preserve the task-relevant information needed for downstream decisions. It also motivated broader questions about why models may perform individual capabilities successfully but struggle when those capabilities must be combined over longer or unfamiliar tasks.

More broadly, I am interested in understanding where and why learned systems fail: whether the required information is missing from their representations, is initially available but lost over time, or is represented internally but not used effectively. I am particularly drawn to research that combines controlled evaluation and failure analysis with methods for improving model behavior through adaptation, training objectives, model architecture, or better interfaces between representations and downstream components.

Research Interests

• Foundation models and autonomous agents
• Representation learning and generative modeling
• Reasoning, sequential decision-making, and compositional generalization
• Model adaptation and post-training
• Rigorous evaluation and failure analysis
• Multimodal and embodied learning

I have previously worked on academic research projects in machine learning and computer vision under the guidance of:

• Prof. Seyda Ertekin at METU, on large language model-guided sequential hyperparameter optimization
• Prof. Ioannis Kakadiaris at the University of Houston, on computer vision and representation learning

Before starting my PhD, I also worked in industry on applied machine learning problems spanning optimization, NLP, computer vision, forecasting, and decision-support systems. These experiences continue to shape my interest in learning systems that are both scientifically grounded and practically useful.

Feel free to explore my Education, Research Experience, Professional Experience, Publications, and CV.