My research focuses on bridging fundamental theory with real-world applications in machine learning and deep learning. I work on a broad spectrum of topics, including but not limited to large language models, generative AI, visual representation learning (deep metric learning), content-based image retrieval, knowledge distillation, learned image compression, recognition/detection, semantic segmentation, multi-target tracking, and stereo matching.

For my full publication list, please visit my Google Scholar Page. My Ph.D. thesis presents mathematical models to explain and optimize deep metric learning (lib.metu) and my M.Sc. thesis explores dynamic programming techniques for fast stereo matching (lib.metu).