Products I've built outside of work, typically to solve a problem for myself or my family.
Penny
Personal finance app · 2026
Live app · sample data
The user problem
Existing budgeting apps offered limited category customization and no view of historical trends. To see how my finances changed MoM and YoY, I tracked everything in spreadsheets: hours of manual work every month.
How it solves it
Penny automates that work. It links real bank and investment accounts through Plaid, sorts transactions into custom categories, shows MoM, YoY and projected spending, and flags categories trending above average.
What I learned
Data synthesis. Aggregating thousands of transactions into views (MoM, YoY, 12-month rolling) that users understand at a glance. User flows. Grouping transactions into trips, recognizing trip spending automatically, and deducting Venmo paybacks from what you paid. Shipped end to end on Vercel and Supabase.
My family began learning MCR mahjong, where a hand needs at least 8 points to win. Scoring is complex, and some players (sorry, Mom!) needed help confirming whether a hand qualified.
How it solves it
Players enter their hand and Pung calculates the score, explains each scoring pattern in English and Chinese, and determines payouts. Hong Kong rules are also supported.
What I learned
Designing for different users. Family members new to MCR shaped bilingual support, plain-language explanations, and a mobile-first flow. Edge cases. The rulebook has many (kongs, last-tile wins, overlapping patterns), so I defined each as a requirement and validated it against real hands.
Small and mid-sized streamers struggle to grow their communities on Twitch.
How it solves it
Live Lounges let streamers co-stream in shared spaces, combining their communities so they can grow together.
What I learned
User empathy. Interviews with 5 streamers surfaced their real pain points, and I iterated until the concept was a frictionless solution they would actually want to use. I then took it from wireframes to a PRD and final pitch.