Education
Publications
-
Learning Relativistic Geodesics and Chaotic Dynamics via Stabilized Lagrangian Neural Networks
Authors: A. U. Hamzaoğulları, A. Ozakin
arXiv preprint • January 2025
We propose Hessian regularization and physics-aware improvements to address training instabilities in Lagrangian Neural Networks. Our methods enable learning complex chaotic systems like triple pendulums and extracting spacetime metrics from geodesic trajectories in general relativistic settings, including AdSâ‚„ spacetime.
Research Experience
Work Experience
Academic Programs & Workshops
Leadership & Teaching Experience
Scholarships & Grants
Turkish Education Association Full Scholarship
Full Coverage • 2014–Present
Full education support from middle school to university.
TÜBİTAK Undergraduate Scholarship
$1700/year • 2022–Present
Granted for excellence in natural sciences.
TÜBİTAK 2224-D
€600/month • June–September 2025
Grant for international scientific research participation.
Erasmus+ Internship Scholarship
€1000/month • June–September 2025
Supported research internship at Radboud University.
Erasmus+ Internship Scholarship
€1000/month • June–September 2024
Supported research internship at Forschungszentrum Jülich.
Cahit Arf Scholarship
$3000/year • 2023–2025
Awarded for top 100 national exam performance.
Technical Skills
Programming Languages
Python, Julia, Java, C
Machine Learning Frameworks
TensorFlow, PyTorch, Flux (Julia)
Tools & Technologies
Git, LaTeX, Jupyter, Linux
Languages
Turkish (Native), English (Fluent), Spanish (Beginner), German (Beginner)
Quantum Computing
Deutsch-Jozsa, Shor's, Grover's algorithms, Quantum cryptography (BB84)
Certifications
Deep Learning & TensorFlow Specializations (deeplearning.ai)