Dr. Barbara

Dr. Barbara

Making Data Matter

Portfolio

Systems That Think Like Humans Think

Applied cognitive science across data products, model evaluation tools, and knowledge graphs. Each build makes complex information easier to see, compare, and act on.

Shipping builds Behaviorally informed UX Evidence you can show

🎯 Currently Shipping

Active projects with live users. Each respects cognitive load and the way people actually scan, sort, and decide.

Fitness Intelligence Platform

Behavioral Patterns

Production classifier that collapses 14 years of workouts into 3 cognitively usable categories.

  • What you'll see: Intensity bands that match how athletes self-label effort when distracted.
  • Tech: Python, AWS Lambda, MySQL, Streamlit.
  • Cognitive insight: The 7±2 rule becomes 3 workable buckets when cognitive load spikes.
Case study → Live demo → GitHub → Documentation →

Convoscope: Multi-Modal AI Comparison

Model Evaluation

Side-by-side AI output comparison that caps choices to working memory limits, not vendor hype.

  • What you'll see: Progressive disclosure that defaults to 3 options, tolerates 5, and gracefully hides the rest.
  • Tech: Streamlit, OpenAI/Anthropic/Google APIs, PostgreSQL.
  • Cognitive insight: Horizontal comparisons reduce cognitive load versus scrolling long chat transcripts.
Case study → Live demo → GitHub → Documentation →

Beehive Knowledge Graph

Metadata Integration

Four years of photos turned into a queryable memory system that mirrors how beekeepers recall outcomes.

  • What you'll see: 7 relationship types that match domain experts' mental models (weather → outcomes).
  • Tech: Neo4j, Google Cloud Vision, Python, Weather APIs.
  • Cognitive insight: Association-first navigation beats chronological browsing for decision recall.
Case study → Live demo → GitHub → Documentation →

🔬 Research & Exploration

Experiments where cognitive science theory meets model behavior and knowledge graphs.

Academic Citation Network Prediction

Knowledge Graph

Predicting missing citations across 8,000 papers to map what researchers notice—and overlook.

  • What you'll see: Attention-like link patterns in scholarly discovery surfaced via graph embeddings.
  • Tech: Graph Neural Networks, PyTorch, Neo4j, Semantic Scholar API.
  • Cognitive insight: Academic attention mirrors visual attention—predictable and bias-prone.
Data story → Live demo → GitHub → Documentation →

🧠 More Builds & Evaluations

Additional systems that translate cognition into better retrieval, summarization, and decision support.

ChronoScope: AI-Powered Timeline Builder

Temporal Extraction

Turns resumes and letters into interactive timelines with confidence scores so dates and events stay trustworthy.

  • What you'll see: Local-first document parsing plus validation that flags uncertain extractions.
  • Tech: Python, Streamlit, PyMuPDF, TimelineJS.
  • Cognitive insight: Time-based narratives help people anchor memory and spot gaps faster.
Case study → GitHub → Documentation →

Poolula Platform: RAG-Powered BI

RAG Evaluation

Natural-language BI with a built-in test harness so answers are measured before they're trusted.

  • What you'll see: Structured SQL answers blended with document citations and accuracy scores.
  • Tech: FastAPI, SQLModel, ChromaDB, Claude.
  • Cognitive insight: Trust emerges when evidence is surfaced alongside the response.
Case study → GitHub →

CareerCraft: Job Application Assistant

Applied GenAI

Guides applicants through tailored cover letters with ATS gap analysis to reduce cognitive overload.

  • What you'll see: Iterative GPT prompts plus keyword scoring before applications go out.
  • Tech: Streamlit, OpenAI API, AWS EC2.
  • Cognitive insight: Progressive disclosure lets people focus on one improvement at a time.
Case study → Live demo → GitHub →

📚 Foundational Builds

Earlier systems that shaped my approach to attention management, progressive disclosure, and information architecture.

IoT Temperature Sensor Fleet (2021)

Streaming Interfaces

Distributed sensor network that revealed how real-time data must be chunked for rapid comprehension.

  • What you'll see: Live readings organized by urgency, not chronology, to reduce alert fatigue.
  • Tech: Microcontrollers, MQTT, Python dashboards.
  • Cognitive insight: Progressive disclosure turns noisy streams into actionable signals.
6-part series →

Digital Portfolio Workshop Series (2024)

Information Architecture

Hands-on sessions that codified a repeatable IA for data portfolios.

  • What you'll see: Reusable IA templates and GitHub Pages workflows that mirror how readers skim.
  • Tech: Jekyll, GitHub Pages, markdown-driven templates.
  • Cognitive insight: Chunked navigation and consistent widths reduce decision fatigue.
Part 1 → Part 2 → Part 3 → Part 4 →

The Cognitive Thread

Technology succeeds when it aligns with how humans naturally think.

Whether it's limiting options to reduce cognitive load, organizing knowledge by association, or chunking live data into usable groups—each build prioritizes how people perceive and decide.

Let's connect →