Data is eating the world.
I'm Rusen, an AI & Data Engineer building tools to make sense of it all.
Tools that transform data into insights.
Classify Anything
ClassificationZero-shot text classification. Define your own classes, paste any text, get predictions.
Segment Anything
SegmentationClick any object in an image to instantly segment it. SAM 2.1 running entirely in your browser via WebAssembly.
Vision Anything
ClassificationZero-shot image classification. Drop an image, type the labels you care about, see how a CLIP-class model ranks them.
See how the data sausage is made.
Emergence
SimulationSimple rules, complex behavior. An interactive essay on cellular automata, synchrony, segregation, highways, and critical cascades.
Embedding Explorer
VisualizationVisualize how texts cluster in vector space using UMAP. See semantic similarity in action.
Sentence Surgeon
InferenceClick any word to remove it. A small BERT predicts what should fill the gap, ranked with probabilities. Click a prediction to graft it in.
Software that lives outside the browser.
Eduport
Local-firstSingle-user desktop app for tracking university applications. Storage is plain Markdown + YAML, sync-friendly and Obsidian-compatible.
vaultdb
QueryingA database engine for your markdown files. Query, filter, mutate, and traverse Obsidian vaults from the command line.
metuclass
Incremental syncCLI tool to sync ODTUClass course files to your local machine. Tracks changes so subsequent syncs only pull new or updated files.
Long-form notes on AI and the practice of programming.
Prompt Is Not Enough: Intent Engineering
ENCan the distance between human intent and model behavior really be closed with prompts alone? In long and risky workflows, what does it take to preserve intent? Let me introduce you to Intent Engineering.
AI Recommendations
ENA guide for aspiring AI researchers and engineers — from finding your motivation to mastering deep learning, computer vision, NLP, and data engineering.
Building In The Age Of AI
ENAs AI evolves, the way digital builders work is changing at breakneck speed. What should we do, how should we adapt, and most importantly — what should we not do?