S — SIDE BUILDS
Built to understand.
These are not case studies. They are experiments — things I built outside of work to test whether an idea was real, to learn a system by building it, or to find where friction actually lives rather than where people assume it does.
Memory Wallet
A design-first take on 'memory you own' — see, redact, and carry what AI products remember about you. Built to test the idea, then sharpened by what the research turned up.
Agent Fleet Console
An attention layer for overseeing many agents at once. A RiskGate triages what genuinely needs a human from what doesn't, so oversight — and trust — scales past a single agent.
Agent Trust Dashboard
A design-first answer to a hard question: what does an AI agent remember about you, and can you take it back? Inspect, redact, and audit agent memory in one surface — with an autonomy dial that turns 'how much should it act on its own' into a control, not a guess.
MyActiveSG+ Rebuild
Rebuilt the booking and balloting flow to understand where the friction actually lives, rather than where the roadmap assumed it was.
Rooster Clinical Monitor
Rebuilt OGP's clinical alert monitor to understand how urgency hierarchies work in healthcare UI — when does an alert need to interrupt, and when does it just add noise?
Service in Motion
A calm, accessible take on order tracking — motion as reassurance on an anxiety surface, then tested for whether it actually lowers the temperature.