Automated first-pass quality audits and surfaced remediation opportunities.
Result
Scoped by shadowing the QA process end to end. A hybrid pass: 17 of 43 criteria resolve deterministically by script, then the remaining 26 go to AI for a verdict with cited evidence. Benchmarked at 100 courses in 50 minutes on a full production cohort.
Draft and transform learning materials into Canvas-ready courses with WCAG 2.1 AA and ASU Design standards embedded.
Result
Course build reduced from 3 to 4 weeks to 2 to 3 days when curriculum is supplied. In a pilot cohort survey, all 8 respondents rated it 4 or 5 out of 5 for overall satisfaction and for improving the speed and quality of their work.
Automated training evaluation reporting and L&D data analysis system.
Result
5 months in production. Survey participation rose from 44.49% to 69.17%, a 56% increase.
An alt-text experiment became a broader page-level accessibility analysis workflow.
Result
Pairs a WCAG script against the markup with a vision pass over a screenshot, catching mismatches neither finds alone, such as alt text that does not describe its image. Code was donated to another high impact project.
A better authoring environment emerged from firsthand friction with live Canvas editing and evolved into organizational tooling.
Result
A separate build-and-review environment designed from interviews with four instructional designers. Its successor, ID Workbench, was merged into the AI Course Authoring System.
An AI-supported de-escalation simulation where trainers practice a difficult conversation instead of reading about one.
Result
Rated 4 out of 5 by the leaders and trainers who used it. A live AI employee pushes back, and the learner passes only after meeting all four de-escalation criteria.
Prompt practice with live grading across five dials, built as the practice half of the prompt engineering pair.
Result
Optional practice, rated 5 out of 5 by the associates who signed up for it. Learners draft a prompt and a live model grades it, rewarding iteration over moving on.
A branching leadership scenario with ten endings, so managers feel the consequence of a call rather than being told it.
Result
One month out, 87% of managers reported more confidence and 100% said the practice method helped them learn.
One tight module taking a beginner from AI terms to functional prompt engineering, scoped to a single outcome.
Result
Post-survey, 80% strongly agreed they understood how AI works and felt more confident using it at work.
MUS 354: The Beatles
With Dr. Christi Jay Wells, consulted 1:1 in an assessment redesign workshop to rework Module 1 and 2 assessments with evidence-based AI-resistant prompting.
IAL 598: The U.S. Intelligence Communities in Practice
With Anthony Hale, retired Deputy Chief of Staff for Intelligence, built an AI-forward intelligence elective.
AI and the Law
With Sean Harrington, Director of the AI and Legal Tech Studio at the Sandra Day O'Connor College of Law, built an AI-forward tech application course.
STP 530: Applied Regression Analysis
With Dr. Yi Zheng, Associate Professor in the School of Mathematical and Statistical Sciences, built a project-based online course using statistics and R.