Geometric research wiki for AI systems, multimodal intelligence, and mathematical structure

Build a field of ideas, not a pile of posts.

I am Brice, an engineer and researcher-in-training working across generative AI, multimodal systems, probability, and performance-aware computing. This site is a connected atlas of notes, derivations, experiments, and essays, shaped as a graph instead of a timeline.

Enter the Atlas About Brice

Focus AI Systems + Multimodal

Method Math, notes, experiments

Format Wiki, graph, pathways

Field Coordinates

Role Engineer + Researcher

Current Base University of Auckland

Primary Vector Generative AI x Multimodal Systems

Mathematical Core Probability, Optimization, Dynamics

Atlas Protocol

graph > chronology structure > feed systems + math + models

Start with one route. Each path opens a cluster of notes, articles, and backlinks inside the atlas.

Route 01

Transformers

architecture, tokenization, scaling, implementation

Route 02

Mathematical Foundations

probability, stochastic processes, optimization, theory

Route 03

AI Systems

inference, memory, infrastructure, performance constraints

Route 04

Master of AI

course notes, linked study paths, architectural reasoning

Manifest

This is where models, mathematics, and systems are allowed to touch each other.

I am most interested in the edges where architectural design, optimization dynamics, uncertainty, and systems constraints stop being separate topics and start behaving like one structure.

Research Vector

• Generative and multimodal AI
• Probability and stochastic processes
• AI systems and inference efficiency
• Computational graphics and geometric thinking

Start Here

Deep learning roadmap Build GPT-2 from scratch Topic hubs and graph routes Soft computing and uncertainty CPU shader experiment Background and direction