Professional Portfolio
Information Systems (Honors) · University of Utah · Graduating May 2027
My work lies at the intersection of business and technology. I use Python, SQL, and Salesforce to turn messy financial data and manual processes into clear, practical tools. Off the clock I'm skiing at Alta, climbing at the Front, mountain biking, or out with my dog, Luna.
Pronto is a food-delivery app, and we focused on the security side: what happens when a customer's account is hacked. My team built an AI chatbot that helps them even though they're locked out, escalates to Trust & Safety and freezes refunds, all in one week on Salesforce.
I was the second tester for an AI support bot a friend built for a crypto derivatives exchange. With an AI browser agent in a live Telegram group, I ran a 32-check test plan covering simultaneous users, human handovers, error screenshots, leaked keys and prompt injection, then wrote up the results.
Is a bigger LLM worth it for cleaning messy financial data? I built a Python pipeline that generates messy transaction data with known answers, cleans it with a rule-based baseline and with LLMs of different sizes, and then scores every run on accuracy and cost in R.
I built this to feature ~170 practice questions verified by running real queries. It includes a lab feature that replays any SELECT clause by clause, letting you literally see how the rows change at each step.
Runs each question's code in a sandbox and explains every miss. My favorite feature is "Dev Mode," which traces the program line by line so you can visualize the call stack and variable history in real time.
I wanted a better way to listen to documents offline, so I built an app that reads PDFs and Word docs aloud with a local neural voice model, highlighting each sentence as it goes. It can also export whole page ranges as audiobooks, generated in parallel across CPU threads.
Smaller builds, mostly tools I made to study for exams.