Sulayman
Yusuf

CS Researcher & ML Engineer

Specializing in Geometric Deep Learning and Mechanistic Interpretability

Research Fellow @ Algoverse AI
Student @ UW Allen School

Loading feature space...

Interactive Feature Space • Sparse Autoencoder Visualization

Current Work @ Algoverse AI

Optimizing ML Training Pipelines

Real-time gradient flow across loss landscapes • 30% faster convergence

Gradient Descent • Loss Landscape Optimization

Impact By The Numbers

0.0
GPA
0+
Model Configurations
0%
Runtime Reduction

Research Interests

Exploring the intersection of geometry, symmetry, and machine learning

SO(3) Equivariance • Manifold Learning • Geometric Symmetry

Geometric Deep Learning

Neural architectures respecting geometric structure of data

Sparse Autoencoders

Interpretable features through geometric sparsity constraints

Equivariant Networks

Models preserving symmetries and transformations in data

Latest Publication

RT-TopKSAE: Improving Top-k Sparse Autoencoders with the Rotation Trick

Under review at ICLR 2026 @ Geometric Representations and Mechanisms (GRaM) Workshop