Technical founder + researcher bridging statistical learning and modern ML

• 📝 Applying to OpenAI Residency 2026 - Research on interpretable AI for marginalized populations
• 📊 Published in Psychological Methods - Co-authored methodology paper on latent transition analysis
• 🛠️ Building research-to-product pipelines - EdTech SaaS translating statistical models into accessible platforms
• 🌍 Studying neurodivergent learning ecosystems - Atlas ERA global worldschooling research network

Statistical Learning → Modern ML:

• Mixture Modeling - Latent class/transition analysis (LCA/LTA) in Python & R
• Modern ML - Neural networks from scratch, transformer architectures, deep learning
• Research Infrastructure - Production LTA pipelines, automated data cleaning, RESTful APIs
• Ethics & AI - Trauma-informed system design, fairness in non-stationary distributions

Tech Stack:

• Languages: Python, R, SQL, JavaScript, TypeScript
• ML/Stats: scikit-learn, statsmodels, pandas, NumPy, TensorFlow (learning)
• Automation: n8n, Airtable API, Google Apps Script
• Infrastructure: Supabase, Firebase, React, Next.js

Production LTA pipeline processing 7,000+ observations with 94% classification accuracy. Custom EM algorithms, RESTful API, automated data cleaning (85% time reduction).

Event-driven workflow automation managing nonprofit operations. Eliminated 15+ hours/week, increased grant compliance from 60% → 100%.

Trauma-informed mental health platform for neurodivergent adolescents. React + Firebase + Supabase, HIPAA-compliant architecture.

NumPy-only implementation demonstrating deep learning mechanics. 92% MNIST accuracy, built from mathematical specifications.

Global worldschooling research network studying neurodivergent learning pathways. Longitudinal measurement frameworks, community-centered research.

Nylund-Gibson, K., et al. (2023). Ten frequently asked questions about latent transition analysis. Psychological Methods, 28(2), 284-300. doi:10.1037/met0000486

Tartt, E. (2023). Unraveling Hopelessness: A Latent Class Analysis of Black Adolescent Student Experiences. Doctoral dissertation, UC Santa Barbara. eScholarship

Interpretable AI for Marginalized Populations

• Fairness in longitudinal models with non-stationary distributions
• Representation learning for small-n heterogeneous subgroups
• Explainable AI for high-stakes educational decisions

Human-Centered ML Evaluation

• Participatory ML where communities define success metrics
• Feedback loops between model predictions and lived experience
• Trauma-informed AI system design

Statistical Learning ↔ Deep Learning Bridges

• Mixture models as structured priors for neural architectures
• Combining interpretable statistical models with flexible neural networks
• Transfer learning for developmental psychology research

PhD in Education - University of California, Santa Barbara (2023) Applied unsupervised learning to identify latent subpopulations in adolescent mental health data. Published in top-tier quantitative journal.

Rapid Learning Velocity:

• Dissertation: 0 → publication in Psychological Methods in 18 months
• ML Self-Study: Statistical methods → transformers in 8 months
• Neural network from scratch achieving 92% MNIST accuracy

Unconventional Path:

• Education research → AI/ML
• Statistical learning → modern deep learning
• Researcher → technical founder/builder

Measurement Ally (For-Profit EdTech SaaS) Founder & CEO - Research-to-product company, IP/code security, micro SaaS development

US-SQUARED (Nonprofit 501c3) Founder & Executive Director - Teen mental health, AI ethics, operations automation programs

• 📧 Email: erica@measurementally.com
• 💼 LinkedIn: linkedin.com/in/ericatartt
• 🐦 Twitter: @ericatartt
• 🌐 Website: ericatartt.com

Your brilliance is not conditional.

Building AI systems that recognize and amplify the strengths of people who've been systematically overlooked by standard training distributions.

Currently: Transitioning from statistical learning to frontier ML research Next: OpenAI Residency 2026 → Interpretable, fair AI for underrepresented populations Mission: Make high-quality research infrastructure accessible to communities who need it most