After years of making data visualizations for major newsrooms and clients, I kept running into the same problem: how do you make sense of information that exists in too many dimensions for humans to naturally understand? These tools are my attempt to build bridges between the abstract mathematical spaces where modern AI lives and the visual, spatial thinking that humans excel at.
What started as a need to understand my own research archive evolved into a suite of tools for mapping everything from film aesthetics to code relationships. Each project explores the intersection of machine learning, interface design, and human cognition—taking complex algorithms and making them visually intuitive and actionable.
Related Content
- Browser Tabs to VR: Creating my Digital Memory Palace - Unity-based 3D visualization of bookmarks with spatial organization
- Turning My Bookmarks Into A Knowledge Graph - Building structured relationship data from web content
Project List
-
criterion-embedding-viz: Vector embeddings for Criterion films enabling searches like "films about existentialism" for cinephiles who want deep thematic connections.
-
code-network-gen: Generate a graph of nodes/edges from your javascript code. Visualizes codebases as interconnected networks treating software architecture as explorable graphs.
-
connectology: Network visualization tool with an 11-page README for those who think in connections and need to see how everything relates to everything else.
-
latent-scope: A scientific instrument for investigating latent spaces. Democratizes advanced ML visualization with extensive documentation making complex embeddings accessible to non-experts.
-
obsidian-analysis: Generate semantic embeddings for your Obsidian vault using LM Studio and Nomic embeddings. Search your notes using natural language and explore with interactive UMAP visualizations.