Ryan 0.00 a curious human

hidden layer 1 · experience

Experimental design for AI features. Voice-first agents. Causal inference at scale. view → Meta 0.00 Data Scientist · now AI-native iOS apps: Waves AI, LOOT Finance, Swole. view → LOOT 0.00 Founder · now Measurement frameworks. Partner analytics. 33% reduction in job failures. view → TikTok 0.00 Data Scientist Built analytics from zero for 1,000+ franchisees. 1% revenue lift. view → Mathnasium 0.00 Sr. Data Scientist Medication adherence. Privacy-preserving analysis. view → SCAN Health 0.00 Data Analyst Machine learning, predictive analytics, big data. view → Pepperdine 0.00 M.S. Analytics Statistics, human behavior, experimental design. view → Arizona State 0.00 B.A. Psychology

hidden layer 2 · essays

Paradoxically, profit will occur when they run out of resources. view → Better Dishwashers 0.00 May 2026 The mirror from the machine. view → Speculum Ex Machina 0.00 June 2025 Why do today what AI can do tomorrow? view → Intelligence Deflation 0.00 April 2025 Starting the conversation. view → The Beginning 0.00 March 2025

hidden layer 3 · products

AI-native personalized meditation built exactly for what you need, when you need it. view → Waves AI 0.00 iOS · meditation Personal budgeting done the right way. No spreadsheets, just info. view → LOOT Finance 0.00 iOS · budgeting AI-native workout tracking and programming. A personal trainer in your pocket. view → Swole 0.00 iOS · fitness

hidden layer 4 · research

15 Kahneman-Tversky biases, 6 frontier models, 4,725 evals per model. Opus 4.6 least susceptible. view → KahneBench 0.00 cognitive biases Do models "think" differently in different languages? 7 cognitive domains, minimal pair design. view → WhorfBench 0.00 linguistic relativity Multi-head attention, transformer blocks, 124M-1.5B parameter configs in PyTorch. view → dummyGPT 0.00 GPT from scratch Evaluation awareness and preference concealment in small models. Published on PyPI. view → Red-teaming 0.00 deceptive alignment Constitutional AI patterns and crisis-response testing for Waves AI. view → meditation-alignment 0.00 production safety

let's talk 0.00 P(we should talk)

model card

Ryan Hartman

v2026.06

Alignment nerd. Data Scientist. Doing useful things.

Riding the shoggoth across Alignment, Data Science, and life in general. Trying to leave the world better than I found it.

base_model: human · mostly aligned
fine_tuned_on: data science · alignment research
deployed_at
location: San Francisco, CA
known_limitations: cannot stop building benchmarks

Full story View my projects Read my essays