For 2 years, from August 2024 to June 2026, I tutored math for the Grossmont-Cuyamaca Community College District. Every month I worked with more than 40 college students on everything from pre-algebra through calculus.
My method was guided questions. I kept asking until I found the exact misconception, then let the student work the problem through on their own. I build software the same way: before I write any code, I find out who is stuck and exactly where.
A lot of what I build started as a problem I ran into myself. My student government tallied office hours by hand, so I built ASGC OS. The board adopted it after one demo, and weekly attendance reporting for our 15 members went from about 5 hours to 10 minutes. California's community college application, CCCApply, once took me 4 Zoom calls to figure out, so at an AI bootcamp I built an assistant for it with 2 teammates. It answers questions in 15 languages and cites its sources.
A good tutor checks the work, and so does my software: AI proposes, and plain code decides. In Pear, Gemini reads the label while a rules engine makes every safety call, and that engine misses zero allergens across more than 1,800 tests. I build with AI coding agents the way a lead runs a team: I write the specs, design the architecture and review every change.
Now I'm studying computer science at San José State University (SJSU) with a 4.0 and a spot on the President's List. I'm looking for a Summer 2027 full-stack internship, and I'm bringing the tutor's habits with me: find the misconception, then show the work.