Research Experience
Oct 2025 - Present
Undergraduate Researcher
University of Central Florida
- Lead a research team of three on deep learning-based Virtual Place Recognition (VPR).
- Developed "PDAL," a novel geometry-preserving loss function, surpassing traditional Triplet Loss by 10x in spatial accuracy (Accepted at IGARSS 2026).
- Engineered a high-efficiency Visual Place Recognition (VPR) framework that matches state-ofthe-art accuracy on bird's-eye view (BEV) datasets while reducing parameters by 45× and embedding size by 64× compared to 1B-parameter models.
- Peer Reviewer for IEEE IROS, IEEE RA-L.
Projects & Publications
PDAL: A GEOMETRY-PRESERVING LOSS FOR CNN-BASED LOCATION RECOGNITION
Celik, F. I., et al. "PDAL: A GEOMETRY-PRESERVING LOSS FOR CNN-BASED LOCATION RECOGNITION" IEEE International Geoscience and Remote Sensing Symposium (IGARSS). Accepted / Expected Publication: October 2026.
AI@UCF Location Recognition
2025 – PresentFounded and led a PyTorch-based project for campus-wide location recognition at UCF.
Awards & Honors
- 1st Place – International High School Capture The Flag Competition (AI & Data Science), UC Santa Barbara, 2023
- 3rd Place – Teknofest A.I. for Health (Technical Lead), 2022
Profile
Specializing in Deep Learning, Computer Vision, and Feature Extraction.
Technical Skills
Python C++ C Java
JavaScript PHP PyTorch
OpenCV TensorFlow ONNX
NumPy Pandas Matplotlib
Education
University of Central Florida
B.S. Computer Science
Expected May 2028
Coursework: OOP, Calculus I-III, Linear Algebra
Certifications
- • Neural Networks & Deep Learning (DeepLearning.AI)
- • CCST Cybersecurity (Cisco)
- • IT Specialist in Java (Certiport)
Languages
English, Turkish