Projects
Things I’ve built.
A mix of data analysis, financial models, machine learning, and product work. Each project includes the process, the result, and what I learned.
My favorites
Rocket Launch Dashboard
Live
I like rockets, so I built a dashboard that tracks launches and other interesting info. A free, automated pipeline pulls launch data from a public API using Python, transforms it with SQL, and updates the dashboard with no server or database.
Independent portfolio project
In the real world
The same fetch, transform, and publish steps run behind most dashboards a data team keeps alive, and this one is also handy for anyone who chases launches and wants to know what's going up next, from where, and how many have flown this year.
Skills shown
- ETL pipelines
- Scheduled automation
- API integration
- Incremental loading
Tools
- Python
- SQL
- DuckDB
- GitHub Actions
- Astro
Chestnuts
I built a fictional, Acorns-inspired investing app, generated its data from scratch, and used it to find where new users get stuck before their first investment.
Independent portfolio project · Fully synthetic data
In the real world
Most of a product analyst's week looks like this, where the team asks why people sign up but never invest, and the job is to turn that into a question the data can answer, find the step where people get stuck, and then test a fix.
Skills shown
- Funnel analysis
- Experiment design
- Data modeling
- Synthetic data
Tools
- Python
- SQL
- DuckDB
- dbt
- Tableau
Exploring Asheville Airbnbs
Using SQL, Excel, and Tableau to shortlist areas for a hypothetical travel company's accommodation partnerships.
Portfolio case study — hypothetical business scenario
In the real world
Travel and partnerships teams do this before anyone picks up the phone, taking a messy public dataset and narrowing a whole city down to the few areas worth calling hosts in, while being upfront about what reviews can and can't say about bookings.
Skills shown
- Data cleaning
- Market comparison
- Dashboard design
Tools
- SQL
- MySQL
- Excel
- Tableau
Other projects
Loan Default Risk AnalysisCredit riskA SQL-first analysis of 601 personal loans to identify the borrower characteristics most associated with default.Aug 2026
CaseTrends: Immigration IntelligenceProduct buildAn end-to-end product build that turned public F2A applicant signals, official Visa Bulletin context, saved-case monitoring, and AI-assisted analysis into a working immigration intelligence tool.Jul 2026
Why We Buy ProductsAI researchI compared how four sets of LLM-generated scores interpreted the same 16,436 Amazon reviews across five types of consumer value—and found that model choice can change the result.Jun 2026
Disney DCF ValuationCorporate financeHow I used Disney’s financial history to build a five-year Excel forecast and estimate the value of its future cash flows.Dec 2025
Predicting Titanic SurvivalMachine learningA compact machine-learning case study comparing logistic regression and a decision tree on the Titanic passenger dataset.Apr 2025
Loan Default Risk Analysis
A SQL-first analysis of 601 personal loans to identify the borrower characteristics most associated with default.
Portfolio case study — educational dataset
In the real world
Credit teams at banks and lenders make this call every day when deciding who gets approved and who needs a closer look, and they lean on the same signals I found here, like credit score, debt-to-income, and how long someone has been working.
Skills shown
- Risk segmentation
- Threshold analysis
- Business recommendations
Tools
- SQL
- MySQL
- Python
- Pandas
CaseTrends: Immigration Intelligence
An end-to-end product build that turned public F2A applicant signals, official Visa Bulletin context, saved-case monitoring, and AI-assisted analysis into a working immigration intelligence tool.
Case study - built, launched, sunset 2026
In the real world
This is what owning a product end to end looks like at a small startup, where you decide what the data is allowed to say, wire up sign-in and billing, keep an AI model honest with guardrails, and then make the hard call on whether to keep running it.
Skills shown
- Product ownership
- AI guardrails
- Data ingestion
Tools
- Next.js
- TypeScript
- Supabase
- Stripe
- Anthropic Claude
- Vercel
- Product Strategy
Why We Buy Products
I compared how four sets of LLM-generated scores interpreted the same 16,436 Amazon reviews across five types of consumer value—and found that model choice can change the result.
Research analysis — 2026
In the real world
Any team using an LLM to tag reviews, support tickets, or survey answers has to ask this before trusting the output, because swapping one model for another can quietly change the conclusions, and checking that up front is cheaper than finding out from a bad decision.
Skills shown
- Model comparison
- Descriptive statistics
- Reproducible research
Tools
- Python
- R
- SQLite
- API Batch Processing
- Excel
Disney DCF Valuation
How I used Disney’s financial history to build a five-year Excel forecast and estimate the value of its future cash flows.
Case study - Fall 2025
In the real world
Equity research and corporate development teams build models like this to decide what a company is really worth, and most of the job is defending each assumption, like using a normal tax rate instead of the one-time negative one Disney reported in 2025.
Skills shown
- Revenue forecasting
- Cost of capital
- Free cash flow
Tools
- Financial Modeling
- Valuation
- DCF
- Excel
Predicting Titanic Survival
A compact machine-learning case study comparing logistic regression and a decision tree on the Titanic passenger dataset.
Mini ML case study - R
In the real world
Swap passengers for customers and survival for churn, fraud, or loan approval, and this is the same workflow companies use for yes-or-no predictions, where the real work is picking a cutoff and explaining what each kind of mistake would cost the business.
Skills shown
- Classification
- Model evaluation
- Data preparation
Tools
- R
- Logistic Regression
- Decision Tree
- caret
- pROC
No projects use that tool yet.