QuintAce Communities
Player-run communities for a poker app: tables, a message board and a weekly leaderboard. An experiment in AI-assisted design, with the PRD and first layouts generated by AI.
PRD to dev-complete on staging in under a week.
I'm Bryan Leung, a senior product designer in the San Francisco Bay Area with more than ten years in B2B SaaS, education technology and consumer games. My work includes Canvas LMS, Zynga Poker and, most recently, AI-assisted product design at QuintAce.
Open to senior product designer and lead roles
5 case studies · 2019 to 2026
Player-run communities for a poker app: tables, a message board and a weekly leaderboard. An experiment in AI-assisted design, with the PRD and first layouts generated by AI.
PRD to dev-complete on staging in under a week.
Tailoring course material to individual students took about 15 clicks per item, so instructors avoided it. I made whole modules assignable.
Ten materials: 150 clicks to 6.
Course Pacing could only set one pace for a whole course. I redesigned it for sections and individual students.
Released to beta with section and student paces.
Instructors published 80 to 120 course items one click at a time. One engineer and I fixed that in a hackathon week.
A 120-item course: 120 clicks to 2.
Players thought they had won hands they lost, and called the game rigged. I redesigned how the table shows a winning hand.
App Store rating back to 4.6. Refund calls down 10%.
Supporting work · 2020 to 2022
I led the experience design team on a match-3 game with about 400k daily players.
Club Challenge shipped on time. Revenue was mixed.
Leagues, Challenges, the VIP Center and the store, during the move from native iOS to Unity.
Leagues raised retention and revenue per player.
San Francisco Bay Area
I design products with complicated rules behind simple screens: course tools for instructors on Canvas LMS, and live games at Zynga and QuintAce. At Zynga I also led the design team on Harry Potter: Puzzles & Spells.
I work in Figma and Unity, and more recently with Claude to get to a testable prototype faster.