Hi, I'm Mason.
I'm a software engineer based in Hong Kong. Outside of software, I spend a lot of time playing basketball, training, learning about AI and markets, and thinking about risk, investing and how to make better decisions.
I spent five years building backend systems at Crypto.com. This site is where I keep track of what I'm learning, thinking about, reading, building and experimenting with over time.
Currently
- AI and stock trading.
- How to avoid losing money, antifragility, and trading psychology.
- Fooled by Randomness — Nassim Nicholas Taleb
- 你快樂所以你成功
- Basketball — training and playing competitive games.
- Gym.
Things I've built
Gomar.ai
An AI gym coach and workout tracker I co-created and launched. It personalises workouts from training history, recovery, goals and equipment, with logging, progress tracking, offline support and a 1,300+ exercise library.
NBA Predictor
A Go and Gin service with PostgreSQL that integrates NBA data behind a layered HTTP API, with Docker, database migrations, YAML configuration and graceful shutdown.
Interests
Basketball
I play regularly and compete in games. It's a big part of how I spend my time outside work.
Gym & fitness
I train regularly and like learning how to improve performance, strength and recovery.
Investing & trading
I'm interested in markets, risk, trading psychology and, increasingly, how to avoid losing rather than only how to make more.
AI & technology
I'm learning how AI is changing the way we build software, work and make decisions.
Building products
I enjoy turning ideas into useful products and learning from actually putting them in front of people.
Work
KYC / EDD Platform
Backend-driven dynamic forms on versioned YAML schemas, supporting regulatory flows across 10+ countries. Rollout workload went from roughly 2.5–4 weeks to 3–5 days.
Crypto.com AI Assistant
A Go feature-eligibility API covering 30+ product features, and a multi-stage intent pipeline where semantic matching handled around 90% of queries in roughly 15–30 ms.
Localization Platform
Localization delivery across 20+ languages, reducing peak infrastructure and translation costs by roughly US$7,000 a month.