Ph.D. · Learning Experience Portfolio
I design learning experiences end-to-end: identifying the opportunity, building the course and the platform it runs on, and measuring whether people learned. For the past four years I've done that in data science, machine learning, and AI education at the University of Arizona. My courses are built in public, revised every term, and designed with AI tools at every stage, from research and prototyping to content and feedback. I'm also a Carpentries Instructor Trainer, so I train other instructors in the same evidence-based methods I use.
01 · Courses
Each course below is a complete educational product, from audience analysis through delivery and revision. Materials are published openly, so the design decisions can be inspected and reused.
Case Study 01 — Teaching beginners
The briefTake graduate students, many arriving from non-technical backgrounds, through the entire data science lifecycle — collection, cleaning, exploration, modeling, communication — in a single semester.
The designPython-first with real-world datasets, so every skill is learned in the context it will be used. Reproducible workflows (Git, Quarto) from week one, and modular AI tutoring from my research — retrieval-augmented feedback, adaptive quizzes, and code-similarity checking.
The evidenceThe full course is open source, with automated GitHub Classroom test suites checking correctness, style, and edge cases — professionals enrolled confirm it matches how they work.
Case Study 02 — Designing where no playbook exists
The briefBuild a graduate deep-learning course in a domain that changes every semester, where proven instructional playbooks don't exist yet.
The designFundamentals-first: backpropagation, optimization, and regularization built up from scratch, so learners have the foundations to follow each new architecture wave. Students build, train, and evaluate real models, with gamified exams and a custom interactive visualization that makes the invisible mechanics visible.
The evidenceVersioned by semester, and powered by an AI Sandbox I fund as Lead-PI of a NAIRR Classroom Pilot — the university's only competitively awarded NAIRR Pilot allocation.
Case Study 03 — Business goals → learning outcomes
The briefIndustry sponsors arrive with business problems. The design task is turning them into scoped, achievable capstones that serve both the sponsor and the learner.
The designI help scope sponsor goals into well-defined projects, then support student teams from scoping through delivery with mentored check-ins and free instructional computing resources.
The evidenceGrew the program from seven students to 100+ — the College's largest applied program — and a $245.5K NSF award I won as Lead-PI let 10+ teams build LLM projects a standard course couldn't run.
02 · AI-First Learning Engineering
I use AI throughout my design work. Prompt engineering, the Claude and OpenAI APIs, RAG, and agentic workflows are everyday tools for research, prototyping, content production, and feedback.
A national-infrastructure sandbox for applied neural-network education, where learners experiment on real GPUs. I designed the sandbox and won the award as Lead-PI.
Sol is an open-source agentic AI with modular tutoring and feedback components, built from my research and piloted across my courses to scale support beyond instructor hours.
A gamified, interactive Jupyter environment for teaching reproducible data science — motivation design and automated checks built directly into the learner's working environment.
Lead education & outreach for MESA, a $4.6M NSF cyberinfrastructure project; supervise an AI + XR/VR experiential program; building EduCloud, provider-agnostic educational computing.
03 · Measurement
04 · Mentoring & Standards