From my MS in AI in Business at ASU W. P. Carey (December 2026), the first AI graduate degree from a U.S. business school.
Every applied project, the method and the question it answered.
| Project | Methods | What it answered |
|---|---|---|
| Underwriting risk model | Regression | How to price risk more accurately |
| Credit risk and churn scoring | Decision tree, random forest, AdaBoost, logistic regression; GridSearchCV; selection by AUC | Which customers are likely to default or leave |
| Loan eligibility app | Classification, deployed with Streamlit | A live web app non-technical users can run themselves |
| Cross-sell propensity | Pruned decision tree; confusion matrix and AUC | Which customers will accept a loan offer |
| Agentic model development | Claude Code, agentic code review, documentation in Linear, Obsidian and GitHub | How to go from a business question to a tested, documented model in one session |
| Language models for gene editing | Evaluation of small language models; research papers | How language models can help design gene-editing enzymes |
| Wearable health patterns | Random forests and neural networks on fitness-tracker data | Which health patterns wearables can reveal for chronic illness management |
| Portfolio segmentation | K-means and hierarchical clustering | How to turn a portfolio into personas executives can act on |
| Customer review sentiment | NLP with VADER, text visualization | What customers are saying, at scale |
| Leadership communication sentiment | NLP with custom metrics for clarity, emotional intelligence and persuasiveness | How effective a team’s written business communication is |
| Political discourse analysis | NLP and sentiment analysis with Voyant | How coverage of an issue shifts across regions and 16 years of debates |
| Interactive dashboards | Tableau, with parameters | How to make data explorable for decision-makers |
| Economic resilience | Data analysis and visualization of GDP, tourism and disaster data | How an economy responds to natural disasters |
| AI tool evaluation | Evaluation framework and presentation | How to judge a new AI tool before adopting it |
| Data governance | NIST and CIA-triad frameworks | How to keep data secure and trustworthy |
From my MS in AI in Business. The first is about governing AI across a company. The other six are machine learning models. For each one I started from a business decision and set the success bar before training. I also compared against a simple baseline and wrote down what the model can’t do.
What I use day to day, grouped by job.
Claude (my main platform), ChatGPT, Gemini, Copilot, open-source and private models, ElevenLabs voice AI, and the ML models I build myself.
Claude Code, Claude skills, MCP, prompt engineering, Make, Superblocks, Lovable, Supabase, ElevenLabs
Python, pandas, scikit-learn, TensorFlow, R, MySQL, Jupyter, Google Colab
Tableau, Streamlit, Vercel, interactive data visualization, AI interface design, anime.js
Linear, Wrike, Jira, Obsidian, GitHub, Excel, AI intake and scoring, portfolio tracking, monthly leadership reporting
NIST AI Risk Management Framework, NIST Generative AI Profile, ISO/IEC 42001, confidentiality, integrity and availability for data, AI risk tiering, vendor AI and data disclosure reviews, AI-agent code review
Executive presentations in PowerPoint, live product demos, change management, AI training and curriculum design
My gene-editing research led to my job at Neoclease.
Papers on how language models can help design gene-editing enzymes.
An NLP model that grades emails and meeting transcripts for clarity and persuasiveness.
Random forests and neural networks on fitness-tracker data, looking for patterns that help manage chronic illness.
Sentiment in 16 years of debates and headlines on immigration, border cities versus the rest.
How Costa Rica’s tourism economy responds to earthquakes, floods and storms.
Master of Science in Artificial Intelligence in Business
BA in Digital Computational Studies · BA in Sociology