Akira
Interdisciplinary researcher at the intersection of mathematics, philosophy, code, and music.
Current Focus
Making AI systems transparent and accessible—creating simple showcases where you can take a small model that fits in laptop memory and understand exactly how it works. Everything open-source, everything designed to democratize understanding.
Research & Building
Transparent AI Systems — Creating browser-based tools that reveal how language models actually work. Using WebGL and open-source models to make weights, activations, and latent spaces visible and interactive.
Human-Centered Interfaces — Designing software that respects bounded human attention. Progressive disclosure, thoughtful interaction patterns, and tools that support flow rather than interrupt it.
Mathematical Pattern Recognition — Applying algebraic geometry and manifold theory to distinguish real mathematical structures from noise in complex systems, from financial time series to neural transformations.
Musical Composition — Organ arrangements exploring how textual themes translate into musical architecture, following the path from narrative to harmonic structure.
Background
- Algebraic Geometry & Manifold Theory — Applied to pattern recognition in complex systems, distinguishing rigid mathematical structures from noise
- NLP for Financial Markets — Built systems to extract trading signals from Israeli business press, analyzing market player ontologies
- Time Series Analysis — Research on prediction methods with focus on making mathematical assumptions explicit for appropriate model selection
- Dynamical Systems Theory — Explored topological transformations in tensor spaces and representation theory
- Philosophy of AI Alignment — Developed gamified systems for value articulation using academic philosophical frameworks
- Musical Composition — Organ arrangements translating narrative motifs into musical structures, recently performed arrangements of poetry
View projects →
Contact
- GitHub: @uprootiny
- Telegram: @uprootiny