Researcher in mathematical foundations of deep learning — spectral methods, geometric ML, and the theory of attention mechanisms.
Currently graduated from B.Sc. in Computer Engineering, where I work on the spectral and geometric structure of self-attention in transformer architectures.
Spectral & Geometric Analysis of Attention Mechanisms
Investigating why attention weights behave the way they do, geometrically
and spectrally. Connects to Teo & Nguyen (NeurIPS 2024).
Distributed ML Infrastructure (TÜBİTAK 2209-B)
Designed and deployed Kolektif, a distributed computing cluster system for ML workloads.
Grammar Evolution & Symbolic Computation
Independent work on evolutionary computation for symbolic regression.
Spectral methods · Kernel theory · Geometric deep learning
Attention interpretability · Mathematical ML
LinkedIn: linkedin.com/in/ruasnv